{ "profile": "l4", "slo": { "ttft_ms": 1000.0, "tpot_ms": 100.0 }, "environment": { "sglang": "0.5.22", "sglang_kernel": "0.4.9", "flashinfer": "0.7.1", "torch": "2.14.1+cu130", "cuda": "13.0", "driver": "615.71.09\n615.71.09", "gpu": "NVIDIA L4", "compute_capability": "8.9", "gpu_memory_total_gib": 22.1, "gpu_memory_free_gib_at_start": 21.9, "python": "3.12.15", "machine": "x86_64", "profile": "l4", "models": [ "Qwen/Qwen3-8B", "Qwen/Qwen3-8B-FP8", "Tengyunw/qwen3_8b_eagle3" ] }, "load": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 13.42, "min": 13.4, "max": 13.6, "n": 3 }, "median_ttft_ms": { "median": 189.16, "min": 184.95, "max": 191.93, "n": 3 }, "p99_ttft_ms": { "median": 696.1, "min": 682.03, "max": 699.33, "n": 3 }, "median_tpot_ms": { "median": 60.16, "min": 60.16, "max": 60.17, "n": 3 }, "p99_tpot_ms": { "median": 103.39, "min": 103.38, "max": 103.41, "n": 3 }, "goodput_rps": { "median": 0.09, "min": 0.08, "max": 0.09, "n": 3 }, "request_throughput": { "median": 0.1, "min": 0.1, "max": 0.11, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1185.0, "max": 1275.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.491 }, "ttft_quantiles_ms": [ 70.8, 70.8, 71.6, 71.9, 162.5, 171.1, 171.4, 175.5, 176.9, 176.9, 177.4, 178.1, 178.2, 178.3, 178.9, 180.8, 181.1, 181.1, 181.3, 182.6, 182.7, 183.4, 184.5, 185.0, 185.4, 185.7, 187.3, 188.6, 193.3, 196.6, 207.8, 228.6, 287.7, 332.8, 332.8, 353.3, 364.7, 424.3, 435.7, 441.4, 495.2, 510.9, 589.6, 590.6, 608.9, 690.5, 693.7, 693.9, 694.9, 696.5, 700.3 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.2, 60.2, 60.2, 60.2, 60.4, 60.4, 60.5, 67.5, 71.8, 71.8, 74.7, 78.4, 79.9, 87.7, 88.2, 90.6, 100.4, 103.3, 103.3, 103.4, 103.4, 103.4, 103.4, 103.4, 103.4, 103.4 ] }, { "x": 2.0, "output_throughput": { "median": 31.2, "min": 31.2, "max": 31.2, "n": 3 }, "median_ttft_ms": { "median": 115.43, "min": 114.91, "max": 115.66, "n": 3 }, "p99_ttft_ms": { "median": 133.75, "min": 132.82, "max": 134.3, "n": 3 }, "median_tpot_ms": { "median": 63.66, "min": 63.65, "max": 63.66, "n": 3 }, "p99_tpot_ms": { "median": 63.88, "min": 63.87, "max": 63.88, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1507.5, "min": 1500.0, "max": 1530.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.4, 96.4, 96.7, 96.7, 96.9, 97.0, 97.1, 97.1, 97.1, 97.2, 97.2, 97.3, 97.3, 97.3, 97.5, 97.5, 97.5, 97.5, 97.5, 97.6, 97.7, 97.8, 97.9, 98.1, 99.1, 99.3, 131.6, 132.0, 132.2, 132.3, 132.3, 132.3, 132.3, 132.3, 132.3, 132.4, 132.5, 132.6, 132.6, 132.6, 132.7, 132.7, 132.8, 133.0, 133.2, 133.2, 133.3, 133.3, 133.4, 133.8, 134.5 ], "tpot_quantiles_ms": [ 63.4, 63.4, 63.4, 63.4, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.7, 63.7, 63.7, 63.7, 63.7, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.9, 63.9, 63.9, 63.9, 63.9, 63.9 ] }, { "x": 4.0, "output_throughput": { "median": 61.54, "min": 58.26, "max": 61.56, "n": 3 }, "median_ttft_ms": { "median": 137.53, "min": 137.09, "max": 505.23, "n": 3 }, "p99_ttft_ms": { "median": 139.33, "min": 139.01, "max": 668.79, "n": 3 }, "median_tpot_ms": { "median": 64.36, "min": 64.34, "max": 64.37, "n": 3 }, "p99_tpot_ms": { "median": 64.8, "min": 64.77, "max": 68.18, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.46, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.46, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1500.0, "max": 1515.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 83.6, 83.6, 84.5, 86.1, 86.3, 86.4, 87.0, 87.0, 87.9, 88.4, 135.0, 135.5, 135.9, 136.3, 136.8, 137.0, 137.0, 137.0, 137.3, 137.5, 137.6, 137.8, 137.9, 138.0, 138.0, 138.1, 138.2, 138.4, 138.5, 138.5, 138.8, 138.8, 138.9, 139.0, 139.0, 139.4, 170.8, 183.4, 193.4, 481.5, 482.0, 482.4, 505.0, 505.5, 618.1, 619.0, 619.6, 664.1, 664.6, 665.1, 669.4 ], "tpot_quantiles_ms": [ 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 65.6, 65.6, 67.4, 67.6, 67.7, 68.3 ] }, { "x": 8.0, "output_throughput": { "median": 112.62, "min": 112.5, "max": 112.67, "n": 3 }, "median_ttft_ms": { "median": 170.43, "min": 169.54, "max": 196.81, "n": 3 }, "p99_ttft_ms": { "median": 1275.48, "min": 1219.4, "max": 1313.98, "n": 3 }, "median_tpot_ms": { "median": 66.01, "min": 66.0, "max": 66.28, "n": 3 }, "p99_tpot_ms": { "median": 74.43, "min": 74.07, "max": 74.54, "n": 3 }, "goodput_rps": { "median": 0.63, "min": 0.6, "max": 0.63, "n": 3 }, "request_throughput": { "median": 0.88, "min": 0.88, "max": 0.88, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1515.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 87.2, 89.8, 90.8, 104.5, 136.5, 137.1, 138.0, 141.1, 141.8, 143.0, 143.9, 155.0, 155.6, 156.0, 156.3, 156.7, 156.9, 157.6, 164.3, 165.2, 166.1, 167.0, 167.5, 168.2, 169.0, 183.9, 191.7, 198.0, 225.3, 497.3, 508.1, 508.9, 621.5, 632.5, 633.6, 732.2, 1198.9, 1199.5, 1204.4, 1205.4, 1206.0, 1206.0, 1206.8, 1207.3, 1217.7, 1218.8, 1219.5, 1275.0, 1275.6, 1313.2, 1314.2 ], "tpot_quantiles_ms": [ 65.8, 65.8, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.1, 66.1, 66.1, 66.2, 66.2, 66.3, 66.3, 66.3, 66.3, 66.3, 66.3, 66.3, 66.4, 66.9, 70.5, 70.5, 70.6, 70.6, 71.3, 71.5, 71.7, 72.7, 74.1, 74.1, 74.7 ] }, { "x": 16.0, "output_throughput": { "median": 201.09, "min": 201.04, "max": 201.65, "n": 3 }, "median_ttft_ms": { "median": 211.78, "min": 211.53, "max": 212.2, "n": 3 }, "p99_ttft_ms": { "median": 2461.58, "min": 2375.12, "max": 2464.91, "n": 3 }, "median_tpot_ms": { "median": 70.11, "min": 70.03, "max": 70.28, "n": 3 }, "p99_tpot_ms": { "median": 87.2, "min": 86.9, "max": 87.39, "n": 3 }, "goodput_rps": { "median": 0.98, "min": 0.93, "max": 1.01, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.57, "max": 1.58, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1425.0, "max": 1455.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 88.9, 92.8, 149.5, 150.6, 151.4, 152.5, 153.7, 155.9, 158.2, 158.7, 159.3, 159.7, 160.1, 161.0, 192.1, 199.5, 200.2, 201.1, 202.2, 202.8, 204.1, 209.9, 210.6, 210.9, 211.3, 211.7, 212.3, 615.7, 619.4, 622.8, 633.5, 779.6, 1521.5, 1781.3, 1782.5, 1784.8, 1797.0, 1799.6, 1803.3, 1804.8, 1823.9, 1824.9, 1970.9, 1972.2, 2346.5, 2362.1, 2374.5, 2400.7, 2459.5, 2461.3, 2465.1 ], "tpot_quantiles_ms": [ 69.6, 69.6, 69.6, 69.6, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.8, 69.8, 69.8, 69.8, 69.8, 69.9, 69.9, 70.0, 70.0, 70.0, 70.0, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.3, 70.3, 70.6, 73.8, 73.8, 73.8, 74.2, 74.2, 74.2, 74.2, 74.3, 74.3, 74.3, 74.4, 77.2, 77.2, 83.2, 83.4, 83.7, 83.7, 86.9, 87.8 ] }, { "x": 32.0, "output_throughput": { "median": 270.11, "min": 269.9, "max": 272.31, "n": 3 }, "median_ttft_ms": { "median": 2969.61, "min": 2908.06, "max": 2983.07, "n": 3 }, "p99_ttft_ms": { "median": 4901.04, "min": 4849.85, "max": 4909.83, "n": 3 }, "median_tpot_ms": { "median": 96.18, "min": 95.91, "max": 96.5, "n": 3 }, "p99_tpot_ms": { "median": 118.02, "min": 117.62, "max": 118.02, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.11, "min": 2.11, "max": 2.13, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1305.0, "max": 1327.5, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 103.8, 198.7, 502.7, 619.1, 627.4, 927.1, 1020.7, 1558.7, 1662.0, 1681.3, 1686.4, 1734.6, 1774.4, 1797.5, 1818.3, 1821.4, 2095.0, 2099.4, 2184.8, 2739.6, 2818.6, 2825.4, 2879.7, 2898.1, 2942.6, 2950.0, 2969.5, 3015.8, 3021.3, 3257.5, 3265.1, 3346.3, 3938.0, 3981.9, 4065.1, 4102.2, 4113.1, 4139.1, 4163.1, 4209.5, 4212.0, 4266.8, 4269.0, 4421.4, 4511.3, 4664.4, 4782.8, 4790.8, 4793.8, 4900.5, 4910.3 ], "tpot_quantiles_ms": [ 81.5, 81.5, 81.7, 81.8, 81.8, 81.8, 81.9, 81.9, 82.1, 83.6, 86.1, 86.3, 86.3, 87.0, 87.1, 87.6, 87.7, 88.1, 89.4, 89.5, 91.2, 92.7, 95.3, 95.6, 95.7, 96.2, 96.5, 96.5, 97.2, 97.3, 97.3, 98.6, 98.9, 100.3, 101.9, 104.5, 105.1, 105.2, 105.4, 105.9, 106.4, 106.8, 106.9, 107.8, 108.2, 109.5, 113.6, 114.6, 115.6, 117.6, 119.0 ] }, { "x": 64.0, "output_throughput": { "median": 250.85, "min": 250.79, "max": 252.85, "n": 3 }, "median_ttft_ms": { "median": 18348.57, "min": 18226.96, "max": 18366.54, "n": 3 }, "p99_ttft_ms": { "median": 33395.92, "min": 33011.34, "max": 33844.62, "n": 3 }, "median_tpot_ms": { "median": 107.42, "min": 107.34, "max": 107.82, "n": 3 }, "p99_tpot_ms": { "median": 163.24, "min": 162.36, "max": 167.66, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.96, "min": 1.96, "max": 1.98, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1080.0, "min": 1020.0, "max": 1200.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.408 }, "ttft_quantiles_ms": [ 182.9, 1682.9, 1813.5, 2827.2, 3171.3, 4148.7, 4675.7, 4761.3, 4791.7, 5298.5, 5407.9, 5488.5, 5531.3, 5612.9, 6976.3, 8249.9, 10726.5, 11172.5, 12080.3, 17185.0, 17197.0, 17314.1, 17345.7, 17458.8, 17681.5, 18228.6, 18357.2, 18427.4, 18512.9, 18646.6, 19373.4, 19514.8, 19550.4, 19646.4, 19712.4, 19833.1, 20721.1, 20929.4, 21071.8, 21438.5, 21787.4, 23224.7, 24176.4, 26479.5, 26692.3, 26914.2, 29161.0, 29511.6, 31847.9, 32767.7, 35160.2 ], "tpot_quantiles_ms": [ 83.7, 83.8, 83.9, 84.0, 84.1, 84.7, 88.3, 89.5, 89.7, 90.3, 90.6, 92.0, 92.7, 94.3, 95.0, 95.4, 97.5, 98.7, 99.0, 99.1, 99.5, 100.0, 101.4, 104.1, 106.9, 107.4, 108.0, 108.5, 108.7, 108.9, 109.6, 111.1, 113.1, 116.4, 117.2, 117.8, 117.8, 118.1, 118.8, 120.1, 120.8, 123.8, 126.0, 134.2, 136.0, 138.4, 144.4, 156.3, 159.5, 162.4, 167.7 ] } ] }, "fp8": { "kv_pool_tokens": 142369, "points": [ { "x": 1.0, "output_throughput": { "median": 24.53, "min": 24.52, "max": 24.59, "n": 3 }, "median_ttft_ms": { "median": 140.72, "min": 139.54, "max": 141.22, "n": 3 }, "p99_ttft_ms": { "median": 255.79, "min": 191.33, "max": 259.14, "n": 3 }, "median_tpot_ms": { "median": 39.66, "min": 39.65, "max": 39.66, "n": 3 }, "p99_tpot_ms": { "median": 43.06, "min": 42.7, "max": 43.23, "n": 3 }, "goodput_rps": { "median": 0.19, "min": 0.19, "max": 0.19, "n": 3 }, "request_throughput": { "median": 0.19, "min": 0.19, "max": 0.19, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1215.0, "min": 1215.0, "max": 1230.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.087 }, "ttft_quantiles_ms": [ 58.9, 58.9, 61.7, 62.9, 124.4, 127.5, 136.0, 136.1, 136.3, 136.6, 136.7, 137.2, 137.4, 138.9, 139.1, 139.1, 139.1, 139.1, 139.1, 139.2, 139.3, 139.4, 139.5, 139.7, 139.9, 140.5, 140.6, 140.8, 141.5, 141.7, 141.8, 141.8, 141.9, 142.2, 142.2, 142.7, 143.1, 143.5, 143.9, 144.0, 145.1, 145.5, 145.9, 146.3, 146.3, 147.9, 149.6, 192.1, 199.3, 271.0, 274.5 ], "tpot_quantiles_ms": [ 39.3, 39.3, 39.3, 39.3, 39.3, 39.3, 39.4, 39.4, 39.4, 39.4, 39.4, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.6, 39.6, 39.6, 39.6, 39.6, 39.6, 39.7, 39.7, 39.7, 39.7, 39.7, 39.7, 39.7, 39.7, 39.8, 39.8, 39.8, 39.8, 39.8, 39.8, 39.9, 39.9, 39.9, 39.9, 40.0, 40.0, 40.3, 40.5, 40.6, 43.1, 43.5, 43.7 ] }, { "x": 2.0, "output_throughput": { "median": 49.96, "min": 49.93, "max": 49.96, "n": 3 }, "median_ttft_ms": { "median": 107.51, "min": 106.99, "max": 108.86, "n": 3 }, "p99_ttft_ms": { "median": 110.09, "min": 110.09, "max": 111.47, "n": 3 }, "median_tpot_ms": { "median": 39.48, "min": 39.47, "max": 39.48, "n": 3 }, "p99_tpot_ms": { "median": 39.57, "min": 39.56, "max": 39.58, "n": 3 }, "goodput_rps": { "median": 0.39, "min": 0.39, "max": 0.39, "n": 3 }, "request_throughput": { "median": 0.39, "min": 0.39, "max": 0.39, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1305.0, "min": 1260.0, "max": 1320.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 103.8, 103.8, 104.8, 105.0, 105.2, 105.2, 105.2, 105.2, 105.4, 105.4, 105.4, 105.6, 105.8, 105.9, 106.2, 106.3, 106.5, 106.5, 106.6, 106.6, 106.9, 106.9, 107.0, 107.2, 108.0, 108.2, 108.3, 108.4, 108.5, 108.5, 108.6, 108.6, 108.8, 108.8, 108.8, 109.1, 109.1, 109.2, 109.2, 109.3, 109.3, 109.4, 109.6, 109.8, 109.8, 109.9, 110.1, 110.2, 110.4, 111.2, 111.5 ], "tpot_quantiles_ms": [ 39.3, 39.3, 39.3, 39.3, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.4, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.5, 39.6, 39.6, 39.6, 39.6, 39.6 ] }, { "x": 4.0, "output_throughput": { "median": 98.8, "min": 98.74, "max": 98.84, "n": 3 }, "median_ttft_ms": { "median": 101.18, "min": 100.04, "max": 103.53, "n": 3 }, "p99_ttft_ms": { "median": 102.56, "min": 102.2, "max": 105.14, "n": 3 }, "median_tpot_ms": { "median": 39.96, "min": 39.96, "max": 39.97, "n": 3 }, "p99_tpot_ms": { "median": 40.18, "min": 40.17, "max": 40.18, "n": 3 }, "goodput_rps": { "median": 0.77, "min": 0.77, "max": 0.77, "n": 3 }, "request_throughput": { "median": 0.77, "min": 0.77, "max": 0.77, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1365.0, "min": 1350.0, "max": 1372.5, "n": 3 }, "max_temp_c": 79.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 75.1, 75.1, 75.6, 75.7, 76.6, 77.0, 77.2, 77.6, 77.8, 78.0, 80.0, 80.4, 80.6, 99.3, 99.5, 99.9, 100.0, 100.0, 100.1, 100.1, 100.3, 100.4, 100.4, 100.5, 100.7, 100.8, 100.8, 100.9, 101.1, 101.1, 101.3, 101.4, 101.4, 101.5, 101.5, 101.6, 101.7, 101.7, 102.0, 102.1, 102.3, 102.6, 103.4, 103.7, 104.1, 104.2, 104.5, 104.7, 104.8, 105.0, 105.2 ], "tpot_quantiles_ms": [ 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 39.9, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.1, 40.1, 40.1, 40.1, 40.1, 40.1, 40.1, 40.2, 40.2, 40.2, 40.2, 40.2 ] }, { "x": 8.0, "output_throughput": { "median": 177.43, "min": 177.36, "max": 178.19, "n": 3 }, "median_ttft_ms": { "median": 128.01, "min": 123.45, "max": 133.89, "n": 3 }, "p99_ttft_ms": { "median": 1055.25, "min": 981.2, "max": 1071.17, "n": 3 }, "median_tpot_ms": { "median": 40.93, "min": 40.91, "max": 40.96, "n": 3 }, "p99_tpot_ms": { "median": 47.91, "min": 47.48, "max": 48.04, "n": 3 }, "goodput_rps": { "median": 1.17, "min": 1.17, "max": 1.39, "n": 3 }, "request_throughput": { "median": 1.39, "min": 1.39, "max": 1.39, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1342.5, "min": 1327.5, "max": 1372.5, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 78.4, 79.5, 79.8, 88.7, 101.5, 101.8, 102.3, 102.5, 102.9, 103.3, 103.7, 103.9, 104.2, 104.5, 104.6, 104.9, 104.9, 105.2, 105.4, 105.7, 106.0, 106.1, 113.9, 114.5, 115.2, 115.8, 149.6, 153.2, 161.3, 391.0, 391.6, 404.7, 507.3, 508.7, 509.5, 518.7, 520.4, 971.0, 972.0, 980.6, 981.3, 984.5, 985.2, 986.3, 998.3, 999.6, 1054.1, 1054.7, 1069.9, 1070.5, 1071.2 ], "tpot_quantiles_ms": [ 40.7, 40.7, 40.7, 40.7, 40.8, 40.8, 40.8, 40.8, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 40.9, 41.0, 41.0, 41.0, 41.0, 41.0, 41.0, 41.1, 41.1, 41.1, 41.1, 44.5, 44.5, 44.6, 45.4, 45.4, 45.6, 45.8, 46.2, 47.5, 47.7, 48.2 ] }, { "x": 16.0, "output_throughput": { "median": 313.92, "min": 313.42, "max": 313.98, "n": 3 }, "median_ttft_ms": { "median": 147.52, "min": 147.51, "max": 148.97, "n": 3 }, "p99_ttft_ms": { "median": 2019.22, "min": 1971.57, "max": 2045.83, "n": 3 }, "median_tpot_ms": { "median": 43.11, "min": 43.07, "max": 43.31, "n": 3 }, "p99_tpot_ms": { "median": 57.29, "min": 56.96, "max": 57.4, "n": 3 }, "goodput_rps": { "median": 1.49, "min": 1.49, "max": 1.53, "n": 3 }, "request_throughput": { "median": 2.45, "min": 2.45, "max": 2.45, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1290.0, "max": 1305.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 80.9, 82.4, 117.0, 118.3, 119.1, 119.9, 120.5, 121.5, 122.6, 123.3, 123.8, 123.9, 124.4, 125.0, 125.2, 125.5, 125.9, 126.2, 126.6, 127.1, 128.0, 137.8, 139.4, 142.6, 146.2, 147.4, 153.3, 386.4, 500.7, 505.9, 509.0, 1373.9, 1374.7, 1376.1, 1376.6, 1484.2, 1485.7, 1487.0, 1488.1, 1489.1, 1492.7, 1494.1, 1505.1, 1506.6, 1832.3, 1942.0, 1946.6, 1971.1, 2017.4, 2019.4, 2046.5 ], "tpot_quantiles_ms": [ 42.9, 42.9, 42.9, 42.9, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.2, 43.3, 43.3, 43.3, 43.3, 43.4, 45.7, 45.7, 45.7, 46.6, 46.6, 46.7, 46.7, 47.4, 47.4, 47.7, 47.7, 48.1, 48.1, 53.5, 54.3, 55.3, 55.4, 57.1, 58.1 ] }, { "x": 32.0, "output_throughput": { "median": 507.2, "min": 505.99, "max": 507.54, "n": 3 }, "median_ttft_ms": { "median": 199.09, "min": 199.04, "max": 201.83, "n": 3 }, "p99_ttft_ms": { "median": 3946.51, "min": 3910.91, "max": 3951.0, "n": 3 }, "median_tpot_ms": { "median": 47.73, "min": 47.71, "max": 47.76, "n": 3 }, "p99_tpot_ms": { "median": 76.1, "min": 76.07, "max": 76.65, "n": 3 }, "goodput_rps": { "median": 2.23, "min": 2.16, "max": 2.23, "n": 3 }, "request_throughput": { "median": 3.96, "min": 3.95, "max": 3.97, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1185.0, "min": 1185.0, "max": 1275.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.7, 123.7, 126.7, 128.2, 129.7, 130.9, 132.4, 133.7, 135.5, 136.5, 154.9, 157.2, 158.7, 159.7, 164.1, 169.5, 170.8, 172.1, 178.4, 180.4, 182.0, 196.9, 197.8, 198.9, 199.5, 200.8, 399.5, 508.8, 1379.7, 1382.7, 1491.1, 1492.8, 1494.4, 1495.8, 1497.1, 2362.7, 2365.5, 2461.1, 2466.9, 2472.8, 2479.3, 3339.4, 3347.2, 3434.6, 3436.5, 3457.2, 3461.0, 3855.2, 3911.0, 3945.5, 3951.4 ], "tpot_quantiles_ms": [ 47.1, 47.1, 47.2, 47.2, 47.2, 47.3, 47.3, 47.3, 47.3, 47.3, 47.3, 47.4, 47.4, 47.4, 47.4, 47.5, 47.5, 47.5, 47.5, 47.6, 47.6, 47.7, 47.7, 47.7, 47.7, 47.7, 47.7, 47.8, 47.8, 50.6, 50.9, 51.0, 51.5, 52.1, 52.1, 58.3, 58.5, 58.7, 59.1, 59.8, 59.8, 66.1, 66.1, 66.2, 66.5, 66.8, 67.5, 67.6, 74.0, 75.3, 77.3 ] }, { "x": 64.0, "output_throughput": { "median": 735.91, "min": 731.55, "max": 739.02, "n": 3 }, "median_ttft_ms": { "median": 296.42, "min": 294.73, "max": 303.46, "n": 3 }, "p99_ttft_ms": { "median": 7903.44, "min": 7740.67, "max": 7940.4, "n": 3 }, "median_tpot_ms": { "median": 56.31, "min": 56.31, "max": 56.33, "n": 3 }, "p99_tpot_ms": { "median": 113.81, "min": 113.39, "max": 114.96, "n": 3 }, "goodput_rps": { "median": 2.87, "min": 2.86, "max": 2.89, "n": 3 }, "request_throughput": { "median": 5.75, "min": 5.72, "max": 5.77, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1185.0, "min": 1185.0, "max": 1192.5, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 108.2, 147.1, 152.9, 163.4, 168.0, 172.4, 181.2, 185.3, 195.2, 199.1, 202.4, 205.4, 209.2, 219.3, 231.7, 244.3, 247.6, 249.9, 252.3, 254.7, 256.8, 259.1, 291.6, 293.8, 296.6, 301.5, 532.8, 1310.3, 1519.3, 1616.0, 2301.1, 2484.3, 2589.9, 3284.2, 3466.7, 3491.9, 3625.5, 4303.3, 4474.4, 4616.5, 5265.5, 5464.1, 5602.7, 6251.9, 6460.6, 6598.4, 7255.5, 7413.4, 7482.9, 7720.4, 7981.7 ], "tpot_quantiles_ms": [ 55.3, 55.3, 55.4, 55.4, 55.5, 55.5, 55.6, 55.7, 55.7, 55.7, 55.8, 55.8, 55.8, 55.8, 55.8, 55.9, 55.9, 55.9, 56.0, 56.1, 56.1, 56.1, 56.2, 56.3, 56.3, 56.3, 56.4, 58.5, 58.6, 59.5, 65.6, 66.5, 67.2, 73.4, 74.4, 75.0, 81.2, 82.1, 82.3, 84.2, 89.3, 90.1, 91.9, 97.0, 97.9, 99.6, 104.7, 105.8, 107.2, 113.4, 117.3 ] } ] }, "dp2": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 11.24, "min": 11.14, "max": 11.31, "n": 3 }, "median_ttft_ms": { "median": 170.82, "min": 170.78, "max": 208.81, "n": 3 }, "p99_ttft_ms": { "median": 352.46, "min": 340.64, "max": 420.73, "n": 3 }, "median_tpot_ms": { "median": 79.59, "min": 79.15, "max": 79.6, "n": 3 }, "p99_tpot_ms": { "median": 145.13, "min": 144.81, "max": 148.9, "n": 3 }, "p99_itl_ms": { "median": 210.19, "min": 175.43, "max": 221.19, "n": 3 }, "median_itl_ms": { "median": 60.35, "min": 60.3, "max": 60.44, "n": 3 }, "goodput_rps": { "median": 0.05, "min": 0.05, "max": 0.06, "n": 3 }, "request_throughput": { "median": 0.09, "min": 0.09, "max": 0.09, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1755.0, "min": 1470.0, "max": 2025.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.196 }, "ttft_quantiles_ms": [ 163.5, 163.5, 163.6, 163.8, 165.4, 165.6, 165.9, 166.1, 166.1, 166.2, 166.3, 166.3, 166.7, 166.7, 166.9, 167.2, 168.4, 168.4, 169.4, 169.4, 169.5, 169.6, 170.1, 170.2, 171.6, 171.9, 172.1, 173.5, 176.0, 189.2, 190.0, 203.0, 203.8, 218.5, 218.5, 227.6, 230.2, 244.5, 245.5, 259.3, 270.9, 289.0, 291.1, 291.8, 297.0, 342.0, 344.8, 349.3, 354.3, 396.6, 425.0 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 98.2, 99.0, 99.1, 99.1, 100.0, 100.4, 100.8, 101.0, 101.0, 101.8, 101.9, 102.3, 102.5, 108.3, 110.4, 112.8, 124.8, 133.0, 133.1, 133.9, 134.3, 142.4, 145.6, 146.7, 151.7 ] }, { "x": 2.0, "output_throughput": { "median": 18.53, "min": 18.42, "max": 20.75, "n": 3 }, "median_ttft_ms": { "median": 241.87, "min": 238.17, "max": 255.32, "n": 3 }, "p99_ttft_ms": { "median": 1237.93, "min": 1088.49, "max": 1336.24, "n": 3 }, "median_tpot_ms": { "median": 79.66, "min": 79.51, "max": 80.62, "n": 3 }, "p99_tpot_ms": { "median": 169.8, "min": 161.86, "max": 190.35, "n": 3 }, "p99_itl_ms": { "median": 223.52, "min": 219.57, "max": 225.38, "n": 3 }, "median_itl_ms": { "median": 60.46, "min": 60.41, "max": 60.72, "n": 3 }, "goodput_rps": { "median": 0.08, "min": 0.08, "max": 0.08, "n": 3 }, "request_throughput": { "median": 0.14, "min": 0.14, "max": 0.16, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1327.5, "max": 1365.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.278 }, "ttft_quantiles_ms": [ 74.3, 74.3, 91.9, 158.0, 168.2, 169.1, 169.6, 169.7, 169.8, 170.6, 170.6, 170.9, 171.1, 171.5, 171.7, 171.8, 172.0, 172.0, 172.5, 172.8, 173.1, 173.2, 173.7, 174.0, 178.9, 182.3, 294.0, 309.7, 331.8, 357.7, 366.2, 376.6, 376.6, 411.7, 411.7, 503.9, 527.7, 540.8, 550.4, 587.3, 620.3, 753.1, 970.1, 1034.3, 1042.3, 1075.7, 1090.7, 1133.9, 1155.3, 1256.3, 1368.2 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 98.9, 99.2, 101.1, 101.4, 103.3, 104.0, 104.2, 105.5, 105.5, 106.0, 107.2, 116.8, 123.4, 126.9, 140.3, 144.5, 148.2, 149.6, 156.0, 157.4, 160.7, 162.6, 167.3, 171.4, 194.4 ] }, { "x": 4.0, "output_throughput": { "median": 39.33, "min": 38.5, "max": 39.86, "n": 3 }, "median_ttft_ms": { "median": 196.81, "min": 182.33, "max": 206.67, "n": 3 }, "p99_ttft_ms": { "median": 566.1, "min": 562.45, "max": 566.71, "n": 3 }, "median_tpot_ms": { "median": 83.91, "min": 83.86, "max": 84.61, "n": 3 }, "p99_tpot_ms": { "median": 117.82, "min": 113.63, "max": 123.76, "n": 3 }, "p99_itl_ms": { "median": 176.52, "min": 164.48, "max": 194.28, "n": 3 }, "median_itl_ms": { "median": 63.73, "min": 63.69, "max": 63.93, "n": 3 }, "goodput_rps": { "median": 0.15, "min": 0.15, "max": 0.16, "n": 3 }, "request_throughput": { "median": 0.31, "min": 0.3, "max": 0.31, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1380.0, "max": 1425.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.078 }, "ttft_quantiles_ms": [ 72.0, 72.0, 72.2, 72.8, 72.8, 75.0, 78.5, 99.9, 102.9, 105.5, 134.3, 135.6, 138.3, 138.9, 139.3, 139.3, 145.4, 145.4, 152.7, 160.7, 169.4, 179.4, 180.2, 180.5, 181.2, 183.5, 184.1, 214.3, 229.2, 348.4, 404.0, 423.7, 474.0, 483.1, 483.1, 485.6, 487.1, 492.1, 494.1, 498.2, 507.6, 510.4, 514.4, 518.7, 526.1, 533.2, 533.8, 545.6, 565.4, 571.9, 575.9 ], "tpot_quantiles_ms": [ 60.1, 60.1, 60.9, 60.9, 61.0, 61.0, 61.0, 61.1, 61.1, 61.4, 61.4, 61.5, 62.2, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.7, 63.7, 63.7, 63.7, 63.9, 63.9, 63.9, 103.8, 104.1, 105.3, 105.9, 106.6, 106.7, 106.8, 107.1, 107.1, 107.6, 108.0, 108.6, 108.9, 109.1, 109.2, 109.2, 109.3, 109.5, 109.7, 109.8, 110.4, 111.1, 114.3, 119.3, 126.0 ] }, { "x": 8.0, "output_throughput": { "median": 64.16, "min": 64.01, "max": 64.19, "n": 3 }, "median_ttft_ms": { "median": 355.73, "min": 347.66, "max": 363.51, "n": 3 }, "p99_ttft_ms": { "median": 13708.74, "min": 13656.37, "max": 15228.45, "n": 3 }, "median_tpot_ms": { "median": 85.96, "min": 85.67, "max": 86.22, "n": 3 }, "p99_tpot_ms": { "median": 142.23, "min": 140.55, "max": 144.48, "n": 3 }, "p99_itl_ms": { "median": 199.21, "min": 167.54, "max": 223.84, "n": 3 }, "median_itl_ms": { "median": 64.62, "min": 64.57, "max": 64.9, "n": 3 }, "goodput_rps": { "median": 0.25, "min": 0.25, "max": 0.25, "n": 3 }, "request_throughput": { "median": 0.5, "min": 0.5, "max": 0.5, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1440.0, "min": 1440.0, "max": 1440.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.136 }, "ttft_quantiles_ms": [ 88.7, 90.4, 101.1, 103.4, 136.4, 138.4, 139.5, 140.2, 145.4, 146.8, 187.1, 188.5, 196.2, 197.7, 237.1, 276.0, 284.1, 290.0, 293.0, 295.6, 296.2, 297.3, 297.9, 298.7, 306.2, 325.7, 401.4, 578.2, 1067.3, 1073.1, 1131.2, 1131.9, 1206.7, 1236.9, 1270.7, 1274.9, 1293.9, 1508.1, 2714.1, 3174.7, 5450.6, 6572.3, 6651.6, 6751.3, 6788.1, 6855.3, 7458.8, 8189.3, 13057.1, 13979.1, 16088.0 ], "tpot_quantiles_ms": [ 60.9, 61.0, 64.2, 64.2, 64.3, 64.3, 64.3, 64.3, 64.3, 64.5, 64.5, 64.5, 64.7, 64.7, 65.1, 65.3, 65.3, 65.4, 65.4, 65.5, 65.6, 65.7, 66.7, 66.8, 66.9, 67.9, 103.4, 104.1, 104.8, 105.6, 105.6, 105.6, 106.1, 107.5, 108.2, 109.2, 109.8, 109.9, 110.1, 110.7, 111.1, 112.6, 112.7, 113.1, 113.8, 131.9, 136.8, 140.5, 142.2, 142.2, 144.5 ] }, { "x": 16.0, "output_throughput": { "median": 50.98, "min": 50.09, "max": 52.25, "n": 3 }, "median_ttft_ms": { "median": 673.05, "min": 620.3, "max": 716.27, "n": 3 }, "p99_ttft_ms": { "median": 72538.8, "min": 69349.42, "max": 74685.38, "n": 3 }, "median_tpot_ms": { "median": 90.82, "min": 88.49, "max": 93.87, "n": 3 }, "p99_tpot_ms": { "median": 175.09, "min": 171.47, "max": 190.6, "n": 3 }, "p99_itl_ms": { "median": 233.36, "min": 232.19, "max": 246.56, "n": 3 }, "median_itl_ms": { "median": 66.25, "min": 65.89, "max": 66.27, "n": 3 }, "goodput_rps": { "median": 0.2, "min": 0.2, "max": 0.2, "n": 3 }, "request_throughput": { "median": 0.4, "min": 0.39, "max": 0.41, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1177.5, "min": 1162.5, "max": 1185.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.41 }, "ttft_quantiles_ms": [ 96.3, 158.2, 165.3, 169.4, 187.0, 207.8, 210.1, 218.1, 244.6, 291.3, 295.0, 299.2, 301.2, 304.0, 308.8, 372.2, 417.7, 427.1, 436.6, 442.5, 447.1, 478.0, 540.7, 586.6, 620.9, 672.6, 717.0, 1132.4, 1246.0, 15454.9, 15597.7, 15949.5, 21403.8, 21517.1, 22260.6, 27226.9, 30407.2, 31555.6, 35002.4, 37752.7, 40424.2, 41655.6, 47365.8, 51821.9, 53294.5, 54986.3, 58120.3, 64318.7, 68139.2, 71398.2, 78329.4 ], "tpot_quantiles_ms": [ 61.2, 63.3, 64.2, 65.3, 65.6, 65.6, 65.7, 65.7, 65.9, 66.0, 66.1, 66.3, 66.4, 66.4, 66.5, 66.9, 67.1, 67.2, 67.3, 67.6, 67.8, 68.0, 68.6, 69.5, 71.7, 81.3, 103.2, 104.3, 104.6, 105.0, 106.1, 106.4, 109.5, 112.7, 112.7, 125.5, 132.0, 140.0, 140.2, 151.4, 158.0, 159.3, 163.4, 163.7, 166.0, 167.2, 168.1, 171.5, 175.1, 178.5, 199.8 ] }, { "x": 32.0, "output_throughput": { "median": 45.21, "min": 45.04, "max": 46.07, "n": 3 }, "median_ttft_ms": { "median": 1616.33, "min": 1497.94, "max": 2091.33, "n": 3 }, "p99_ttft_ms": { "median": 178233.99, "min": 172971.56, "max": 180727.7, "n": 3 }, "median_tpot_ms": { "median": 104.72, "min": 103.54, "max": 105.47, "n": 3 }, "p99_tpot_ms": { "median": 181.53, "min": 179.27, "max": 184.53, "n": 3 }, "p99_itl_ms": { "median": 244.21, "min": 232.22, "max": 254.24, "n": 3 }, "median_itl_ms": { "median": 90.72, "min": 70.11, "max": 96.11, "n": 3 }, "goodput_rps": { "median": 0.1, "min": 0.09, "max": 0.1, "n": 3 }, "request_throughput": { "median": 0.35, "min": 0.35, "max": 0.36, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 765.0, "min": 720.0, "max": 780.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.617 }, "ttft_quantiles_ms": [ 111.4, 203.2, 294.0, 300.8, 305.0, 366.4, 393.0, 436.9, 445.3, 461.2, 509.0, 560.4, 624.5, 692.2, 712.2, 734.7, 1044.5, 1048.9, 1206.9, 1211.5, 1225.1, 1330.9, 1464.7, 1514.1, 1540.5, 1615.9, 2090.9, 15565.8, 15750.4, 30322.3, 46416.5, 47730.5, 57469.1, 76462.3, 82119.6, 97511.5, 103650.4, 119302.7, 131192.3, 136198.4, 146896.1, 149311.7, 151461.9, 154936.0, 161639.5, 164005.2, 167283.6, 169263.6, 171515.5, 176638.6, 181035.0 ], "tpot_quantiles_ms": [ 61.0, 64.3, 65.1, 65.6, 66.1, 67.4, 67.9, 69.1, 69.2, 69.2, 69.3, 69.5, 69.7, 69.7, 71.4, 71.9, 74.0, 74.2, 74.8, 76.0, 77.5, 79.4, 83.2, 89.4, 100.4, 104.1, 105.4, 106.9, 109.2, 109.7, 125.8, 147.0, 152.1, 154.6, 155.3, 159.7, 159.7, 160.9, 162.8, 164.0, 164.4, 165.6, 167.0, 169.3, 171.7, 172.5, 173.7, 174.3, 179.0, 181.5, 184.5 ] }, { "x": 64.0, "output_throughput": { "median": 41.7, "min": 41.29, "max": 42.1, "n": 3 }, "median_ttft_ms": { "median": 4763.42, "min": 4691.37, "max": 4812.79, "n": 3 }, "p99_ttft_ms": { "median": 391308.51, "min": 387144.31, "max": 396672.24, "n": 3 }, "median_tpot_ms": { "median": 109.72, "min": 108.86, "max": 110.68, "n": 3 }, "p99_tpot_ms": { "median": 206.07, "min": 196.39, "max": 206.77, "n": 3 }, "p99_itl_ms": { "median": 249.64, "min": 249.48, "max": 253.31, "n": 3 }, "median_itl_ms": { "median": 90.86, "min": 88.15, "max": 103.26, "n": 3 }, "goodput_rps": { "median": 0.06, "min": 0.06, "max": 0.07, "n": 3 }, "request_throughput": { "median": 0.33, "min": 0.32, "max": 0.33, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 675.0, "min": 585.0, "max": 705.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.671 }, "ttft_quantiles_ms": [ 183.7, 292.5, 305.4, 438.8, 475.4, 538.2, 563.4, 572.7, 658.0, 695.2, 732.9, 778.2, 837.2, 1165.0, 1232.2, 1280.9, 1326.1, 1471.3, 1584.8, 1732.9, 1973.1, 2458.7, 2701.3, 3694.7, 3877.3, 4762.3, 15726.1, 30475.1, 48280.6, 86528.4, 115124.7, 140549.0, 158184.7, 201109.2, 228736.3, 253277.7, 289322.0, 310057.7, 334793.2, 359134.1, 363101.9, 367427.7, 370167.8, 376478.0, 377955.3, 380502.8, 382787.7, 384716.4, 387202.3, 391886.5, 397460.3 ], "tpot_quantiles_ms": [ 60.4, 65.3, 66.3, 67.7, 69.5, 70.3, 74.0, 77.5, 78.5, 80.8, 83.3, 84.1, 84.3, 85.3, 87.9, 90.6, 91.5, 92.7, 94.9, 98.0, 100.6, 102.2, 103.6, 107.2, 108.8, 109.7, 111.1, 118.8, 150.4, 158.4, 161.5, 164.5, 165.4, 167.3, 168.5, 169.6, 170.1, 170.6, 172.7, 174.1, 175.2, 177.0, 177.7, 179.7, 182.6, 186.2, 187.1, 191.5, 195.2, 200.5, 221.7 ] } ] }, "eagle3": { "kv_pool_tokens": 9951, "points": [ { "x": 1.0, "output_throughput": { "median": 31.81, "min": 31.67, "max": 37.05, "n": 3 }, "median_ttft_ms": { "median": 251.91, "min": 236.88, "max": 254.61, "n": 3 }, "p99_ttft_ms": { "median": 260.92, "min": 260.22, "max": 263.44, "n": 3 }, "median_tpot_ms": { "median": 25.61, "min": 22.72, "max": 28.48, "n": 3 }, "p99_tpot_ms": { "median": 58.51, "min": 40.75, "max": 60.31, "n": 3 }, "p99_itl_ms": { "median": 73.44, "min": 73.41, "max": 73.48, "n": 3 }, "median_itl_ms": { "median": 18.34, "min": 18.31, "max": 18.37, "n": 3 }, "goodput_rps": { "median": 0.25, "min": 0.25, "max": 0.29, "n": 3 }, "request_throughput": { "median": 0.25, "min": 0.25, "max": 0.29, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1515.0, "max": 1530.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.73, "min": 2.5, "max": 2.78, "n": 3 }, "ttft_quantiles_ms": [ 85.6, 85.6, 86.9, 87.0, 178.3, 179.6, 187.3, 188.9, 192.8, 194.2, 194.6, 195.0, 196.2, 196.8, 198.0, 198.9, 204.4, 204.4, 222.6, 242.3, 244.2, 246.7, 248.8, 251.2, 251.8, 252.1, 252.7, 253.1, 253.3, 253.5, 253.9, 254.2, 254.2, 255.0, 255.0, 255.5, 255.5, 256.1, 256.1, 256.6, 256.8, 257.2, 258.3, 258.4, 259.0, 259.2, 259.5, 260.6, 260.7, 261.2, 263.9 ], "tpot_quantiles_ms": [ 18.4, 18.4, 19.0, 19.6, 19.6, 20.1, 20.7, 20.7, 20.7, 20.7, 20.8, 20.8, 21.3, 21.3, 21.9, 21.9, 22.4, 22.4, 23.1, 23.1, 23.1, 23.5, 23.6, 23.6, 24.8, 25.3, 25.4, 25.9, 25.9, 27.0, 27.1, 27.6, 28.8, 28.8, 28.8, 29.4, 29.4, 29.4, 31.1, 32.8, 34.5, 34.6, 36.2, 36.3, 36.9, 39.7, 40.3, 40.8, 53.6, 59.4, 63.9 ] }, { "x": 2.0, "output_throughput": { "median": 53.89, "min": 53.57, "max": 67.11, "n": 3 }, "median_ttft_ms": { "median": 326.64, "min": 325.9, "max": 327.47, "n": 3 }, "p99_ttft_ms": { "median": 341.96, "min": 337.08, "max": 379.92, "n": 3 }, "median_tpot_ms": { "median": 28.12, "min": 25.17, "max": 28.43, "n": 3 }, "p99_tpot_ms": { "median": 59.15, "min": 40.71, "max": 60.61, "n": 3 }, "p99_itl_ms": { "median": 75.77, "min": 74.92, "max": 80.29, "n": 3 }, "median_itl_ms": { "median": 18.7, "min": 18.66, "max": 18.76, "n": 3 }, "goodput_rps": { "median": 0.42, "min": 0.42, "max": 0.52, "n": 3 }, "request_throughput": { "median": 0.42, "min": 0.42, "max": 0.52, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1560.0, "min": 1560.0, "max": 1575.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.73, "min": 2.49, "max": 2.77, "n": 3 }, "ttft_quantiles_ms": [ 216.7, 216.7, 237.3, 238.2, 240.0, 248.2, 250.7, 252.1, 313.3, 314.0, 319.7, 320.3, 321.1, 321.6, 322.4, 322.5, 323.1, 323.1, 325.4, 325.5, 325.7, 325.9, 326.0, 326.2, 326.3, 326.7, 327.1, 327.1, 327.6, 327.6, 327.8, 327.9, 328.0, 328.3, 328.3, 328.4, 329.7, 330.1, 330.3, 330.4, 330.4, 330.5, 330.8, 330.9, 330.9, 331.4, 331.7, 335.3, 338.0, 343.9, 387.8 ], "tpot_quantiles_ms": [ 18.9, 18.9, 20.5, 20.8, 21.1, 21.1, 21.1, 21.2, 21.4, 21.5, 21.8, 22.3, 22.5, 22.5, 23.1, 23.7, 23.8, 23.8, 24.4, 24.9, 24.9, 25.4, 25.5, 25.5, 25.6, 27.2, 27.3, 27.3, 27.8, 28.4, 29.6, 29.6, 30.2, 31.2, 31.2, 31.2, 31.3, 31.3, 31.4, 32.5, 34.9, 36.0, 38.4, 39.8, 40.4, 40.7, 40.7, 41.9, 57.5, 59.4, 63.9 ] }, { "x": 4.0, "output_throughput": { "median": 94.18, "min": 86.98, "max": 119.55, "n": 3 }, "median_ttft_ms": { "median": 279.66, "min": 242.37, "max": 318.39, "n": 3 }, "p99_ttft_ms": { "median": 381.88, "min": 345.07, "max": 382.6, "n": 3 }, "median_tpot_ms": { "median": 30.23, "min": 25.93, "max": 32.41, "n": 3 }, "p99_tpot_ms": { "median": 54.14, "min": 44.08, "max": 60.05, "n": 3 }, "p99_itl_ms": { "median": 90.19, "min": 81.19, "max": 106.01, "n": 3 }, "median_itl_ms": { "median": 19.55, "min": 19.15, "max": 19.62, "n": 3 }, "goodput_rps": { "median": 0.74, "min": 0.68, "max": 0.93, "n": 3 }, "request_throughput": { "median": 0.74, "min": 0.68, "max": 0.93, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1620.0, "min": 1537.5, "max": 2025.0, "n": 3 }, "max_temp_c": 80.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.73, "min": 2.48, "max": 2.76, "n": 3 }, "ttft_quantiles_ms": [ 152.9, 152.9, 217.5, 217.7, 218.1, 218.2, 218.2, 218.4, 218.4, 218.5, 218.6, 218.6, 218.7, 218.7, 218.7, 218.7, 218.9, 218.9, 219.0, 219.2, 220.8, 264.7, 265.7, 267.3, 278.9, 279.5, 279.7, 279.9, 280.2, 280.6, 312.6, 324.1, 326.2, 328.4, 328.4, 329.3, 329.4, 329.7, 331.1, 331.2, 332.4, 332.6, 334.3, 334.8, 344.6, 345.2, 349.5, 349.7, 350.1, 388.3, 390.2 ], "tpot_quantiles_ms": [ 19.3, 19.3, 21.0, 21.0, 21.1, 21.5, 21.7, 22.1, 22.4, 22.7, 22.7, 23.6, 24.4, 24.6, 24.6, 24.9, 25.3, 25.3, 25.4, 25.5, 25.7, 26.2, 26.4, 28.4, 28.7, 29.9, 30.1, 30.3, 31.4, 31.6, 32.5, 33.2, 33.6, 33.9, 33.9, 34.7, 36.1, 36.7, 36.7, 36.8, 38.3, 38.6, 40.4, 43.3, 43.3, 44.2, 44.2, 44.2, 55.9, 59.3, 60.2 ] }, { "x": 8.0, "output_throughput": { "median": 181.55, "min": 165.97, "max": 193.53, "n": 3 }, "median_ttft_ms": { "median": 311.23, "min": 291.75, "max": 344.57, "n": 3 }, "p99_ttft_ms": { "median": 524.49, "min": 516.93, "max": 658.33, "n": 3 }, "median_tpot_ms": { "median": 35.58, "min": 32.49, "max": 36.01, "n": 3 }, "p99_tpot_ms": { "median": 87.69, "min": 48.82, "max": 91.47, "n": 3 }, "p99_itl_ms": { "median": 143.63, "min": 141.47, "max": 244.65, "n": 3 }, "median_itl_ms": { "median": 21.54, "min": 21.44, "max": 23.85, "n": 3 }, "goodput_rps": { "median": 1.42, "min": 1.26, "max": 1.51, "n": 3 }, "request_throughput": { "median": 1.42, "min": 1.3, "max": 1.51, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1567.5, "min": 1522.5, "max": 1665.0, "n": 3 }, "max_temp_c": 80.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.74, "min": 2.59, "max": 2.74, "n": 3 }, "ttft_quantiles_ms": [ 159.7, 218.5, 219.1, 234.0, 234.9, 235.4, 235.4, 235.8, 236.3, 269.7, 275.8, 276.6, 282.4, 283.6, 284.1, 284.5, 284.8, 285.2, 285.6, 286.2, 291.6, 291.9, 297.2, 297.8, 298.3, 337.3, 341.3, 342.8, 343.6, 344.9, 346.4, 348.5, 349.0, 349.4, 349.8, 351.2, 351.6, 363.2, 372.0, 403.2, 434.9, 437.2, 449.9, 454.8, 476.0, 488.8, 507.4, 521.2, 522.4, 590.9, 658.7 ], "tpot_quantiles_ms": [ 21.3, 21.6, 23.4, 24.1, 24.2, 24.9, 25.2, 25.8, 27.2, 27.8, 27.9, 28.0, 28.8, 29.0, 29.5, 29.7, 29.7, 30.0, 30.0, 31.0, 31.4, 32.1, 32.6, 33.1, 33.5, 34.5, 35.1, 35.7, 36.0, 36.7, 36.9, 37.3, 37.6, 38.3, 39.6, 39.8, 41.1, 41.4, 41.9, 42.4, 45.0, 45.3, 46.1, 46.5, 47.0, 49.6, 50.0, 51.5, 55.0, 78.7, 100.4 ] }, { "x": 16.0, "output_throughput": { "median": 255.93, "min": 252.28, "max": 263.0, "n": 3 }, "median_ttft_ms": { "median": 949.6, "min": 910.55, "max": 1051.02, "n": 3 }, "p99_ttft_ms": { "median": 3514.28, "min": 3333.85, "max": 3693.25, "n": 3 }, "median_tpot_ms": { "median": 43.59, "min": 43.24, "max": 44.48, "n": 3 }, "p99_tpot_ms": { "median": 105.79, "min": 75.83, "max": 107.08, "n": 3 }, "p99_itl_ms": { "median": 257.54, "min": 257.02, "max": 260.29, "n": 3 }, "median_itl_ms": { "median": 27.24, "min": 27.19, "max": 29.25, "n": 3 }, "goodput_rps": { "median": 1.11, "min": 0.94, "max": 1.12, "n": 3 }, "request_throughput": { "median": 2.0, "min": 1.97, "max": 2.05, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1395.0, "min": 1380.0, "max": 1395.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.74, "min": 2.73, "max": 2.74, "n": 3 }, "ttft_quantiles_ms": [ 382.3, 440.5, 457.7, 492.4, 500.6, 512.3, 570.1, 587.5, 610.1, 621.3, 623.1, 627.2, 629.8, 647.5, 653.4, 663.3, 707.6, 721.3, 733.3, 780.2, 793.6, 809.5, 896.0, 900.9, 924.7, 952.0, 996.1, 1016.4, 1044.4, 1063.8, 1089.6, 1123.3, 1151.6, 1188.6, 1230.1, 1262.2, 1415.8, 1624.5, 1902.2, 2112.0, 2113.4, 2126.6, 2128.0, 2130.9, 2132.2, 2164.7, 2166.0, 2169.9, 2234.4, 2235.7, 6174.5 ], "tpot_quantiles_ms": [ 23.8, 24.6, 26.1, 27.6, 28.9, 30.7, 31.8, 33.5, 34.1, 35.1, 35.8, 36.4, 36.5, 37.3, 38.3, 38.5, 38.9, 39.5, 40.0, 40.4, 40.7, 41.3, 42.2, 43.1, 43.5, 43.5, 43.7, 44.3, 44.5, 45.0, 45.6, 46.4, 48.1, 49.7, 51.0, 52.7, 53.4, 54.9, 56.8, 58.0, 59.3, 60.2, 62.2, 63.6, 68.1, 70.2, 71.3, 73.1, 76.0, 82.2, 119.7 ] }, { "x": 32.0, "output_throughput": { "median": 269.42, "min": 268.6, "max": 269.59, "n": 3 }, "median_ttft_ms": { "median": 8040.24, "min": 8004.07, "max": 8147.79, "n": 3 }, "p99_ttft_ms": { "median": 11791.07, "min": 11335.51, "max": 11871.6, "n": 3 }, "median_tpot_ms": { "median": 43.82, "min": 43.33, "max": 44.92, "n": 3 }, "p99_tpot_ms": { "median": 96.12, "min": 95.62, "max": 99.7, "n": 3 }, "p99_itl_ms": { "median": 259.68, "min": 259.06, "max": 260.58, "n": 3 }, "median_itl_ms": { "median": 28.2, "min": 27.65, "max": 28.6, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.1, "min": 2.1, "max": 2.11, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1305.0, "max": 1327.5, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "accept_length": { "median": 2.74, "min": 2.73, "max": 2.77, "n": 3 }, "ttft_quantiles_ms": [ 1220.0, 2120.4, 2123.7, 2143.0, 2160.3, 2163.5, 5404.6, 6064.5, 6858.4, 7158.7, 7326.0, 7404.9, 7516.3, 7577.2, 7609.4, 7657.4, 7687.1, 7724.9, 7754.6, 7807.0, 7852.7, 7888.6, 7928.0, 7951.8, 7994.7, 8052.2, 8093.6, 8121.4, 8152.2, 8188.7, 8232.7, 8289.2, 8356.6, 8370.2, 8394.4, 8458.1, 8491.0, 8535.0, 8583.6, 8643.7, 8681.4, 8740.7, 8799.6, 8920.1, 8978.3, 9076.4, 9252.0, 9426.3, 9961.1, 11141.9, 12648.6 ], "tpot_quantiles_ms": [ 22.8, 26.1, 30.3, 31.9, 33.6, 34.6, 35.0, 35.7, 36.6, 37.0, 37.4, 37.8, 38.6, 39.0, 39.4, 39.8, 40.4, 40.7, 41.2, 41.5, 41.9, 42.4, 43.0, 43.4, 43.7, 43.9, 44.6, 45.1, 45.8, 46.7, 47.8, 49.2, 50.1, 51.0, 51.9, 53.8, 54.9, 56.8, 59.0, 59.8, 61.1, 63.0, 64.0, 66.1, 68.6, 70.9, 72.9, 76.4, 79.0, 85.7, 115.9 ] }, { "x": 64.0, "output_throughput": { "median": 245.34, "min": 240.43, "max": 250.09, "n": 3 }, "median_ttft_ms": { "median": 23264.21, "min": 23185.63, "max": 23449.15, "n": 3 }, "p99_ttft_ms": { "median": 32375.32, "min": 32274.15, "max": 33884.5, "n": 3 }, "median_tpot_ms": { "median": 51.5, "min": 50.12, "max": 52.2, "n": 3 }, "p99_tpot_ms": { "median": 113.23, "min": 103.69, "max": 114.91, "n": 3 }, "p99_itl_ms": { "median": 363.33, "min": 340.6, "max": 376.7, "n": 3 }, "median_itl_ms": { "median": 30.25, "min": 30.11, "max": 30.33, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.92, "min": 1.88, "max": 1.95, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1095.0, "min": 1080.0, "max": 1117.5, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.361 }, "accept_length": { "median": 2.74, "min": 2.73, "max": 2.79, "n": 3 }, "ttft_quantiles_ms": [ 1247.0, 2146.2, 2182.1, 5665.5, 7054.0, 8713.1, 11902.5, 13815.2, 15924.7, 19204.1, 21094.4, 21798.0, 22006.2, 22171.9, 22306.9, 22406.3, 22501.7, 22562.3, 22664.4, 22740.3, 22825.5, 22944.9, 23043.4, 23130.7, 23189.2, 23269.7, 23345.0, 23416.7, 23512.8, 23607.2, 23784.8, 23901.1, 24095.3, 24352.4, 24668.9, 25073.8, 25634.3, 26020.7, 26306.8, 27120.9, 27833.5, 28527.2, 29157.2, 29738.0, 30245.8, 30754.4, 31061.8, 31529.3, 32219.2, 32900.0, 34699.7 ], "tpot_quantiles_ms": [ 23.9, 28.4, 34.1, 35.2, 36.2, 37.4, 38.1, 38.6, 39.0, 39.7, 40.1, 40.7, 41.5, 41.9, 42.8, 43.5, 43.9, 44.6, 45.3, 46.2, 46.9, 47.8, 48.8, 49.5, 50.1, 51.2, 52.3, 53.6, 54.9, 55.8, 56.8, 58.3, 59.2, 60.1, 61.4, 62.6, 63.6, 65.4, 66.6, 68.1, 69.8, 72.2, 75.2, 77.0, 78.9, 82.2, 86.7, 92.0, 97.5, 104.9, 158.3 ] } ] }, "sglang-h2h": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 13.84, "min": 13.65, "max": 14.07, "n": 3 }, "median_ttft_ms": { "median": 184.79, "min": 181.36, "max": 186.46, "n": 3 }, "p99_ttft_ms": { "median": 621.85, "min": 565.56, "max": 662.01, "n": 3 }, "median_tpot_ms": { "median": 60.16, "min": 60.15, "max": 60.16, "n": 3 }, "p99_tpot_ms": { "median": 101.76, "min": 101.64, "max": 103.31, "n": 3 }, "p99_itl_ms": { "median": 103.53, "min": 103.5, "max": 103.83, "n": 3 }, "median_itl_ms": { "median": 60.2, "min": 60.2, "max": 60.21, "n": 3 }, "goodput_rps": { "median": 0.1, "min": 0.08, "max": 0.1, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.11, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1275.0, "max": 1290.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.434 }, "ttft_quantiles_ms": [ 71.0, 71.0, 71.5, 71.9, 170.0, 170.4, 173.9, 174.1, 176.0, 178.1, 178.2, 178.5, 179.5, 179.7, 179.7, 180.3, 180.6, 180.6, 180.7, 180.9, 180.9, 181.3, 181.4, 182.2, 183.6, 183.6, 183.8, 184.7, 186.0, 186.7, 188.3, 209.7, 292.1, 303.5, 303.5, 308.7, 374.1, 390.6, 454.1, 469.8, 470.1, 476.7, 499.4, 515.9, 523.3, 534.9, 543.1, 560.7, 566.4, 639.2, 694.7 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.5, 60.6, 60.6, 60.6, 69.9, 83.4, 84.0, 84.8, 90.5, 90.7, 90.9, 91.7, 92.6, 103.1, 103.3, 103.3, 103.3, 103.4, 103.4 ] }, { "x": 2.0, "output_throughput": { "median": 31.2, "min": 31.2, "max": 31.2, "n": 3 }, "median_ttft_ms": { "median": 115.74, "min": 115.34, "max": 117.34, "n": 3 }, "p99_ttft_ms": { "median": 135.16, "min": 134.03, "max": 136.48, "n": 3 }, "median_tpot_ms": { "median": 63.65, "min": 63.65, "max": 63.66, "n": 3 }, "p99_tpot_ms": { "median": 63.87, "min": 63.86, "max": 63.87, "n": 3 }, "p99_itl_ms": { "median": 65.08, "min": 64.92, "max": 65.27, "n": 3 }, "median_itl_ms": { "median": 63.52, "min": 63.5, "max": 63.52, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1515.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.3, 96.3, 97.0, 97.0, 97.0, 97.0, 97.3, 97.3, 97.9, 98.0, 98.0, 98.0, 98.1, 98.1, 98.1, 98.2, 98.6, 98.6, 98.6, 98.8, 98.8, 99.0, 99.1, 99.3, 99.3, 102.7, 131.6, 131.8, 132.0, 132.2, 132.2, 132.3, 132.5, 132.7, 132.7, 132.8, 132.9, 133.0, 133.0, 133.2, 133.3, 133.3, 133.4, 133.6, 133.7, 133.7, 134.0, 134.0, 134.6, 135.4, 136.8 ], "tpot_quantiles_ms": [ 63.4, 63.4, 63.4, 63.4, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.7, 63.7, 63.7, 63.7, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.9, 63.9, 63.9, 63.9, 63.9 ] }, { "x": 4.0, "output_throughput": { "median": 61.55, "min": 61.53, "max": 61.56, "n": 3 }, "median_ttft_ms": { "median": 137.27, "min": 137.27, "max": 138.38, "n": 3 }, "p99_ttft_ms": { "median": 138.42, "min": 138.31, "max": 140.63, "n": 3 }, "median_tpot_ms": { "median": 64.36, "min": 64.33, "max": 64.37, "n": 3 }, "p99_tpot_ms": { "median": 64.78, "min": 64.75, "max": 64.79, "n": 3 }, "p99_itl_ms": { "median": 67.06, "min": 64.97, "max": 67.1, "n": 3 }, "median_itl_ms": { "median": 64.33, "min": 64.3, "max": 64.33, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1530.0, "min": 1515.0, "max": 1575.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.2, 86.2, 86.5, 86.5, 86.7, 86.8, 86.9, 87.2, 87.2, 87.8, 87.9, 88.1, 89.3, 136.7, 136.7, 136.7, 137.0, 137.0, 137.1, 137.1, 137.2, 137.2, 137.3, 137.4, 137.6, 137.7, 137.7, 137.7, 137.7, 137.7, 137.7, 137.8, 137.9, 137.9, 137.9, 137.9, 138.0, 138.1, 138.2, 138.2, 138.3, 138.3, 138.4, 138.5, 138.6, 138.6, 139.0, 139.2, 139.3, 140.1, 140.7 ], "tpot_quantiles_ms": [ 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.8 ] }, { "x": 8.0, "output_throughput": { "median": 112.46, "min": 112.38, "max": 112.58, "n": 3 }, "median_ttft_ms": { "median": 169.98, "min": 160.27, "max": 200.89, "n": 3 }, "p99_ttft_ms": { "median": 1277.35, "min": 1230.77, "max": 1365.42, "n": 3 }, "median_tpot_ms": { "median": 66.09, "min": 66.01, "max": 66.32, "n": 3 }, "p99_tpot_ms": { "median": 74.63, "min": 74.26, "max": 74.74, "n": 3 }, "p99_itl_ms": { "median": 70.77, "min": 70.35, "max": 73.3, "n": 3 }, "median_itl_ms": { "median": 65.99, "min": 65.93, "max": 66.04, "n": 3 }, "goodput_rps": { "median": 0.63, "min": 0.6, "max": 0.66, "n": 3 }, "request_throughput": { "median": 0.88, "min": 0.88, "max": 0.88, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.3, 88.0, 89.0, 101.4, 137.6, 138.6, 139.0, 139.3, 139.7, 140.0, 140.6, 141.0, 141.6, 142.3, 143.0, 153.9, 154.4, 154.7, 155.0, 155.4, 155.8, 162.5, 163.4, 164.4, 173.8, 184.3, 196.3, 203.9, 219.4, 488.6, 504.3, 504.3, 529.9, 624.2, 625.0, 803.2, 805.3, 1205.7, 1206.3, 1211.3, 1212.3, 1212.9, 1217.0, 1218.0, 1219.3, 1230.6, 1276.4, 1277.1, 1363.8, 1364.7, 1365.6 ], "tpot_quantiles_ms": [ 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.1, 66.1, 66.1, 66.1, 66.1, 66.3, 66.3, 66.3, 66.3, 66.3, 66.3, 66.4, 66.4, 66.4, 66.5, 66.9, 69.5, 69.5, 71.4, 71.4, 71.6, 71.7, 71.7, 72.6, 74.2, 74.3, 75.0 ] }, { "x": 16.0, "output_throughput": { "median": 201.13, "min": 201.09, "max": 201.21, "n": 3 }, "median_ttft_ms": { "median": 211.91, "min": 211.27, "max": 213.69, "n": 3 }, "p99_ttft_ms": { "median": 2425.75, "min": 2412.93, "max": 2466.53, "n": 3 }, "median_tpot_ms": { "median": 70.06, "min": 70.02, "max": 70.37, "n": 3 }, "p99_tpot_ms": { "median": 87.28, "min": 87.14, "max": 87.31, "n": 3 }, "p99_itl_ms": { "median": 81.41, "min": 80.67, "max": 82.32, "n": 3 }, "median_itl_ms": { "median": 69.64, "min": 69.59, "max": 69.77, "n": 3 }, "goodput_rps": { "median": 0.96, "min": 0.96, "max": 0.98, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.57, "max": 1.57, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1440.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 89.8, 92.7, 150.2, 151.9, 152.9, 154.1, 155.4, 157.6, 159.0, 159.5, 159.9, 160.3, 160.9, 188.3, 194.5, 201.3, 201.9, 202.6, 203.4, 204.0, 206.0, 210.1, 210.7, 211.4, 211.7, 212.4, 213.5, 490.1, 612.3, 621.3, 624.4, 1653.8, 1655.7, 1678.4, 1679.9, 1777.2, 1778.7, 1788.0, 1790.4, 1791.6, 1809.5, 1810.7, 1818.7, 1820.6, 2354.6, 2384.4, 2398.9, 2412.1, 2425.4, 2464.7, 2466.7 ], "tpot_quantiles_ms": [ 69.6, 69.6, 69.6, 69.6, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.8, 69.8, 69.8, 69.8, 69.8, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.4, 70.4, 70.6, 70.9, 74.2, 74.3, 74.3, 74.3, 74.4, 74.4, 75.1, 75.1, 75.6, 75.6, 76.2, 76.2, 83.5, 83.6, 84.2, 85.0, 87.0, 87.8 ] }, { "x": 32.0, "output_throughput": { "median": 271.28, "min": 270.26, "max": 272.19, "n": 3 }, "median_ttft_ms": { "median": 2870.77, "min": 2823.43, "max": 2873.49, "n": 3 }, "p99_ttft_ms": { "median": 4782.15, "min": 4776.84, "max": 4893.23, "n": 3 }, "median_tpot_ms": { "median": 96.79, "min": 96.48, "max": 96.82, "n": 3 }, "p99_tpot_ms": { "median": 117.84, "min": 117.6, "max": 117.85, "n": 3 }, "p99_itl_ms": { "median": 102.06, "min": 101.77, "max": 102.57, "n": 3 }, "median_itl_ms": { "median": 81.57, "min": 81.53, "max": 81.62, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.12, "min": 2.11, "max": 2.13, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1320.0, "max": 1320.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 106.1, 199.8, 495.4, 508.7, 625.9, 1094.7, 1098.1, 1447.7, 1648.3, 1652.0, 1664.8, 1679.7, 1681.9, 1685.1, 1688.7, 1778.3, 1811.0, 2260.2, 2261.4, 2643.4, 2801.4, 2806.8, 2822.9, 2861.6, 2865.3, 2867.5, 2870.4, 2873.6, 2942.2, 3005.6, 3408.4, 3419.6, 3862.9, 3967.3, 3971.3, 3993.7, 4048.5, 4051.0, 4057.1, 4062.4, 4107.4, 4195.1, 4198.3, 4488.6, 4674.4, 4759.0, 4770.1, 4772.0, 4780.5, 4839.0, 4894.5 ], "tpot_quantiles_ms": [ 81.6, 81.6, 81.7, 81.8, 81.8, 81.8, 81.9, 81.9, 82.6, 82.8, 86.3, 87.0, 87.3, 87.4, 87.5, 87.5, 88.1, 88.5, 88.5, 90.1, 91.8, 91.8, 92.1, 95.7, 96.2, 96.5, 96.8, 96.8, 96.9, 97.3, 97.9, 99.1, 99.2, 100.9, 101.1, 105.1, 105.3, 105.6, 106.2, 106.2, 106.2, 106.4, 107.2, 108.2, 108.6, 110.0, 110.2, 114.5, 115.5, 117.8, 119.0 ] }, { "x": 64.0, "output_throughput": { "median": 256.62, "min": 255.76, "max": 260.48, "n": 3 }, "median_ttft_ms": { "median": 18355.96, "min": 18291.44, "max": 18393.38, "n": 3 }, "p99_ttft_ms": { "median": 31321.78, "min": 30933.0, "max": 32221.24, "n": 3 }, "median_tpot_ms": { "median": 106.17, "min": 106.09, "max": 107.03, "n": 3 }, "p99_tpot_ms": { "median": 159.34, "min": 152.5, "max": 161.07, "n": 3 }, "p99_itl_ms": { "median": 114.31, "min": 114.09, "max": 117.69, "n": 3 }, "median_itl_ms": { "median": 83.82, "min": 83.73, "max": 83.85, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.0, "min": 2.0, "max": 2.03, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1215.0, "min": 1200.0, "max": 1215.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.344 }, "ttft_quantiles_ms": [ 197.1, 1836.1, 1841.4, 3006.2, 3025.0, 4177.3, 4667.3, 4767.1, 4775.4, 5345.6, 5437.3, 5476.4, 5497.5, 5597.4, 6520.7, 7594.2, 10300.4, 10667.0, 11618.3, 17192.1, 17275.0, 17299.1, 17309.8, 17435.6, 17506.6, 18292.9, 18394.6, 18463.2, 18493.6, 18613.5, 18995.3, 19481.2, 19544.7, 19652.7, 19693.2, 19796.3, 20548.1, 20669.5, 20691.1, 20856.5, 20918.4, 22631.3, 23200.6, 24401.3, 25815.9, 25926.0, 27066.7, 28080.6, 29734.2, 31075.8, 33619.1 ], "tpot_quantiles_ms": [ 83.7, 83.9, 83.9, 84.0, 84.1, 84.3, 84.7, 85.2, 89.6, 89.7, 90.3, 90.3, 90.7, 92.7, 93.4, 93.9, 96.9, 96.9, 99.0, 99.1, 99.1, 99.5, 100.0, 101.9, 103.1, 106.2, 107.1, 108.4, 108.5, 108.6, 109.3, 111.1, 111.8, 112.9, 115.4, 117.7, 117.7, 117.8, 118.2, 118.7, 119.6, 120.2, 124.0, 130.5, 132.2, 133.0, 138.8, 148.2, 152.5, 159.3, 161.1 ] } ] }, "vllm": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 13.51, "min": 13.47, "max": 13.53, "n": 3 }, "median_ttft_ms": { "median": 191.84, "min": 191.43, "max": 198.04, "n": 3 }, "p99_ttft_ms": { "median": 704.56, "min": 694.91, "max": 705.01, "n": 3 }, "median_tpot_ms": { "median": 60.36, "min": 60.35, "max": 60.36, "n": 3 }, "p99_tpot_ms": { "median": 104.16, "min": 104.15, "max": 104.17, "n": 3 }, "p99_itl_ms": { "median": 104.54, "min": 104.4, "max": 104.71, "n": 3 }, "median_itl_ms": { "median": 60.47, "min": 60.46, "max": 60.52, "n": 3 }, "goodput_rps": { "median": 0.09, "min": 0.09, "max": 0.09, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.11, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1260.0, "max": 1290.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.48 }, "ttft_quantiles_ms": [ 75.9, 75.9, 76.7, 78.1, 174.3, 175.3, 178.3, 178.3, 180.3, 180.4, 183.0, 183.1, 183.4, 185.6, 185.8, 186.3, 186.7, 186.7, 187.0, 187.8, 188.3, 188.4, 189.2, 191.1, 191.3, 191.7, 192.4, 192.6, 198.3, 203.5, 221.7, 223.9, 315.2, 321.4, 321.4, 372.9, 386.6, 389.8, 429.3, 503.9, 512.1, 513.5, 530.4, 538.2, 624.5, 702.8, 703.6, 703.7, 704.7, 705.3, 707.3 ], "tpot_quantiles_ms": [ 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.4, 60.4, 60.4, 60.4, 60.7, 60.7, 62.3, 66.4, 71.4, 71.4, 72.0, 72.3, 86.7, 87.3, 88.4, 92.0, 93.7, 95.0, 98.0, 104.1, 104.1, 104.1, 104.2, 104.2, 104.2, 104.2 ] }, { "x": 2.0, "output_throughput": { "median": 31.26, "min": 31.25, "max": 31.26, "n": 3 }, "median_ttft_ms": { "median": 132.1, "min": 132.01, "max": 132.11, "n": 3 }, "p99_ttft_ms": { "median": 147.73, "min": 143.31, "max": 148.19, "n": 3 }, "median_tpot_ms": { "median": 63.43, "min": 63.43, "max": 63.44, "n": 3 }, "p99_tpot_ms": { "median": 63.55, "min": 63.53, "max": 63.56, "n": 3 }, "p99_itl_ms": { "median": 64.3, "min": 64.26, "max": 64.45, "n": 3 }, "median_itl_ms": { "median": 63.42, "min": 63.42, "max": 63.43, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 2040.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.083 }, "ttft_quantiles_ms": [ 79.1, 79.1, 79.7, 80.0, 128.2, 128.5, 128.6, 128.6, 128.6, 128.7, 128.8, 128.8, 128.8, 128.9, 128.9, 128.9, 128.9, 128.9, 129.0, 129.3, 129.3, 129.3, 129.4, 129.6, 129.7, 129.8, 134.4, 134.4, 134.5, 134.5, 134.6, 135.0, 135.1, 135.5, 135.5, 137.9, 139.2, 139.3, 139.4, 139.4, 139.4, 139.4, 139.5, 139.6, 139.7, 139.8, 140.2, 140.6, 144.0, 149.0, 149.7 ], "tpot_quantiles_ms": [ 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.4, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.6, 63.6 ] }, { "x": 4.0, "output_throughput": { "median": 61.8, "min": 61.79, "max": 61.8, "n": 3 }, "median_ttft_ms": { "median": 140.91, "min": 140.77, "max": 141.61, "n": 3 }, "p99_ttft_ms": { "median": 154.31, "min": 153.14, "max": 154.68, "n": 3 }, "median_tpot_ms": { "median": 64.1, "min": 64.09, "max": 64.12, "n": 3 }, "p99_tpot_ms": { "median": 64.19, "min": 64.18, "max": 64.2, "n": 3 }, "p99_itl_ms": { "median": 64.97, "min": 64.88, "max": 65.19, "n": 3 }, "median_itl_ms": { "median": 64.08, "min": 64.08, "max": 64.11, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1515.0, "max": 1522.5, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 81.5, 81.5, 82.9, 83.2, 132.1, 132.3, 132.8, 132.8, 132.9, 132.9, 133.2, 133.3, 133.3, 139.9, 140.2, 140.3, 140.4, 140.4, 140.6, 140.7, 140.7, 140.8, 140.8, 140.8, 140.9, 141.0, 141.0, 141.1, 141.2, 141.3, 141.4, 141.4, 141.5, 141.6, 141.6, 141.7, 141.9, 141.9, 142.0, 142.5, 143.0, 143.6, 152.3, 152.8, 153.2, 153.4, 153.8, 154.0, 154.3, 154.4, 154.7 ], "tpot_quantiles_ms": [ 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2 ] }, { "x": 8.0, "output_throughput": { "median": 112.79, "min": 112.77, "max": 112.86, "n": 3 }, "median_ttft_ms": { "median": 252.91, "min": 220.96, "max": 259.64, "n": 3 }, "p99_ttft_ms": { "median": 1318.5, "min": 1313.64, "max": 1337.93, "n": 3 }, "median_tpot_ms": { "median": 65.75, "min": 65.62, "max": 65.76, "n": 3 }, "p99_tpot_ms": { "median": 73.43, "min": 73.27, "max": 73.83, "n": 3 }, "p99_itl_ms": { "median": 67.62, "min": 67.58, "max": 70.34, "n": 3 }, "median_itl_ms": { "median": 65.7, "min": 65.62, "max": 65.74, "n": 3 }, "goodput_rps": { "median": 0.72, "min": 0.66, "max": 0.72, "n": 3 }, "request_throughput": { "median": 0.88, "min": 0.88, "max": 0.88, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.5, 86.6, 133.4, 134.8, 138.0, 139.9, 144.4, 145.2, 146.6, 148.6, 162.5, 162.5, 162.8, 163.1, 163.4, 163.7, 164.2, 165.7, 166.2, 166.8, 167.5, 210.1, 210.7, 211.4, 212.8, 214.1, 294.8, 309.4, 320.6, 469.0, 715.7, 716.5, 806.3, 835.3, 836.1, 836.7, 837.8, 868.4, 869.3, 975.8, 1117.2, 1117.6, 1200.7, 1201.5, 1257.4, 1306.2, 1307.2, 1312.6, 1313.9, 1337.2, 1338.1 ], "tpot_quantiles_ms": [ 65.4, 65.4, 65.4, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.6, 65.6, 65.6, 65.6, 65.6, 65.6, 65.7, 65.7, 65.9, 65.9, 65.9, 65.9, 65.9, 66.1, 66.1, 66.9, 68.0, 68.1, 68.8, 68.8, 68.8, 68.8, 69.4, 70.0, 70.0, 70.0, 70.0, 72.2, 73.2, 73.3, 74.2 ] }, { "x": 16.0, "output_throughput": { "median": 200.9, "min": 200.51, "max": 201.21, "n": 3 }, "median_ttft_ms": { "median": 288.72, "min": 281.85, "max": 293.73, "n": 3 }, "p99_ttft_ms": { "median": 2515.04, "min": 2489.52, "max": 2543.37, "n": 3 }, "median_tpot_ms": { "median": 69.5, "min": 69.44, "max": 69.92, "n": 3 }, "p99_tpot_ms": { "median": 85.85, "min": 85.54, "max": 85.96, "n": 3 }, "p99_itl_ms": { "median": 80.14, "min": 79.5, "max": 104.71, "n": 3 }, "median_itl_ms": { "median": 69.47, "min": 69.41, "max": 69.64, "n": 3 }, "goodput_rps": { "median": 1.01, "min": 1.0, "max": 1.03, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.57, "max": 1.57, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1425.0, "min": 1425.0, "max": 1440.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.4, 173.8, 175.1, 180.1, 181.5, 185.4, 186.2, 187.7, 202.1, 205.6, 211.0, 215.3, 216.2, 217.9, 222.5, 231.5, 239.4, 240.6, 241.5, 243.7, 252.0, 253.3, 263.9, 270.1, 271.2, 272.4, 352.7, 839.9, 854.6, 856.9, 880.8, 883.0, 900.5, 1308.9, 1310.3, 1374.2, 1437.4, 1466.4, 1693.3, 1695.5, 1697.8, 1701.3, 1738.6, 1775.4, 1998.3, 2000.8, 2033.8, 2300.8, 2489.9, 2515.5, 2543.6 ], "tpot_quantiles_ms": [ 69.2, 69.2, 69.2, 69.3, 69.3, 69.3, 69.3, 69.3, 69.3, 69.3, 69.4, 69.4, 69.4, 69.4, 69.4, 69.5, 69.5, 69.5, 69.5, 69.5, 69.5, 69.6, 69.6, 69.6, 69.7, 69.7, 69.9, 69.9, 69.9, 73.7, 73.8, 74.2, 74.2, 74.3, 74.3, 74.9, 75.0, 75.1, 77.3, 77.3, 77.5, 78.1, 78.4, 81.3, 81.5, 81.5, 81.6, 81.7, 81.8, 85.8, 86.3 ] }, { "x": 32.0, "output_throughput": { "median": 273.93, "min": 273.81, "max": 274.38, "n": 3 }, "median_ttft_ms": { "median": 1813.49, "min": 1775.41, "max": 1833.2, "n": 3 }, "p99_ttft_ms": { "median": 4839.48, "min": 4822.54, "max": 4886.34, "n": 3 }, "median_tpot_ms": { "median": 102.18, "min": 102.12, "max": 102.26, "n": 3 }, "p99_tpot_ms": { "median": 113.77, "min": 113.46, "max": 114.39, "n": 3 }, "p99_itl_ms": { "median": 618.75, "min": 612.12, "max": 622.46, "n": 3 }, "median_itl_ms": { "median": 80.59, "min": 80.57, "max": 80.62, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.14, "min": 2.14, "max": 2.14, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1260.0, "max": 1290.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 120.6, 394.1, 677.5, 815.2, 833.2, 839.5, 859.2, 1275.9, 1281.8, 1284.4, 1343.9, 1394.7, 1537.0, 1612.5, 1635.9, 1669.4, 1716.8, 1733.9, 1737.9, 1768.1, 1772.8, 1776.9, 1785.8, 1799.9, 1811.9, 1823.1, 1825.5, 1829.2, 1832.6, 1836.2, 1867.3, 1869.8, 1878.9, 1883.5, 1888.6, 1906.2, 1957.7, 2167.9, 2331.4, 2381.3, 2426.8, 2443.1, 2480.1, 2499.5, 3076.5, 3099.2, 3691.9, 4292.2, 4307.7, 4823.1, 4886.8 ], "tpot_quantiles_ms": [ 80.2, 80.3, 82.7, 87.1, 87.3, 91.4, 95.1, 96.2, 100.1, 100.8, 100.9, 100.9, 100.9, 101.1, 101.4, 101.5, 101.6, 101.6, 101.7, 101.7, 101.7, 101.8, 101.8, 101.9, 102.1, 102.2, 102.3, 102.3, 102.4, 102.4, 102.5, 102.7, 103.2, 103.4, 103.6, 104.0, 104.2, 104.5, 104.6, 105.0, 105.2, 105.8, 107.1, 107.9, 108.7, 108.8, 110.0, 111.3, 111.6, 112.9, 115.8 ] }, { "x": 64.0, "output_throughput": { "median": 241.39, "min": 241.37, "max": 246.4, "n": 3 }, "median_ttft_ms": { "median": 10953.28, "min": 10725.3, "max": 11528.47, "n": 3 }, "p99_ttft_ms": { "median": 26231.01, "min": 24116.5, "max": 27502.42, "n": 3 }, "median_tpot_ms": { "median": 129.19, "min": 129.09, "max": 129.78, "n": 3 }, "p99_tpot_ms": { "median": 257.79, "min": 253.99, "max": 260.18, "n": 3 }, "p99_itl_ms": { "median": 1362.94, "min": 1204.34, "max": 1436.32, "n": 3 }, "median_itl_ms": { "median": 83.78, "min": 83.63, "max": 83.89, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.89, "min": 1.89, "max": 1.92, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1012.5, "min": 990.0, "max": 1020.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.449 }, "ttft_quantiles_ms": [ 166.2, 1320.5, 1897.0, 2841.8, 3456.7, 4257.5, 4851.7, 4973.4, 5056.3, 5176.4, 5341.2, 5461.7, 5483.8, 5556.8, 5597.9, 5727.9, 5782.2, 5866.8, 6084.4, 6660.5, 7872.0, 8939.8, 9454.5, 9916.7, 10334.3, 10787.4, 11688.9, 12075.6, 12370.4, 13849.1, 15848.0, 16068.4, 16461.1, 16575.8, 16640.2, 16695.1, 16795.7, 16883.4, 18025.0, 18633.2, 19005.0, 19271.2, 19703.1, 20236.6, 20916.2, 22655.7, 23398.2, 23808.4, 24391.4, 25051.1, 27526.0 ], "tpot_quantiles_ms": [ 71.9, 80.6, 119.8, 119.9, 120.5, 120.7, 120.7, 121.1, 121.3, 121.4, 121.5, 121.6, 121.7, 121.8, 122.0, 122.1, 122.2, 122.5, 122.7, 122.7, 123.0, 123.8, 124.7, 125.6, 127.7, 129.2, 132.3, 138.6, 141.8, 144.5, 146.6, 150.8, 154.1, 159.4, 162.6, 164.7, 165.8, 168.6, 170.7, 171.7, 173.1, 174.1, 176.4, 180.2, 183.7, 186.4, 191.2, 195.1, 204.7, 241.0, 305.8 ] } ] }, "triton": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 13.94, "min": 13.9, "max": 16.36, "n": 3 }, "median_ttft_ms": { "median": 182.59, "min": 181.09, "max": 186.41, "n": 3 }, "p99_ttft_ms": { "median": 677.41, "min": 276.62, "max": 697.72, "n": 3 }, "median_tpot_ms": { "median": 59.98, "min": 59.91, "max": 59.99, "n": 3 }, "p99_tpot_ms": { "median": 100.82, "min": 62.95, "max": 102.03, "n": 3 }, "p99_itl_ms": { "median": 102.19, "min": 61.64, "max": 102.29, "n": 3 }, "median_itl_ms": { "median": 60.03, "min": 59.9, "max": 60.04, "n": 3 }, "goodput_rps": { "median": 0.1, "min": 0.1, "max": 0.13, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.13, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1275.0, "max": 1365.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.435 }, "ttft_quantiles_ms": [ 71.2, 71.2, 71.6, 72.2, 162.1, 169.4, 171.3, 172.9, 173.0, 174.0, 175.9, 176.7, 177.7, 178.3, 179.9, 179.9, 180.0, 180.0, 180.6, 180.6, 181.1, 181.4, 181.5, 181.8, 182.1, 182.3, 182.9, 183.1, 184.4, 184.5, 186.3, 186.5, 187.5, 202.4, 202.4, 203.4, 208.5, 215.2, 288.6, 296.5, 404.2, 469.4, 470.2, 480.7, 480.9, 485.2, 487.3, 502.2, 587.2, 708.3, 717.2 ], "tpot_quantiles_ms": [ 59.8, 59.8, 59.8, 59.8, 59.8, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 59.9, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 61.4, 63.5, 66.4, 83.3, 84.3, 85.0, 86.5, 87.2, 88.5, 94.1, 102.0, 102.0, 102.0 ] }, { "x": 2.0, "output_throughput": { "median": 31.32, "min": 31.32, "max": 31.33, "n": 3 }, "median_ttft_ms": { "median": 104.93, "min": 103.76, "max": 105.14, "n": 3 }, "p99_ttft_ms": { "median": 137.42, "min": 135.74, "max": 137.6, "n": 3 }, "median_tpot_ms": { "median": 63.48, "min": 63.47, "max": 63.49, "n": 3 }, "p99_tpot_ms": { "median": 63.8, "min": 63.78, "max": 63.8, "n": 3 }, "p99_itl_ms": { "median": 64.85, "min": 64.81, "max": 65.09, "n": 3 }, "median_itl_ms": { "median": 63.24, "min": 63.23, "max": 63.24, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1515.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 72.0, 72.0, 72.2, 72.2, 72.4, 72.5, 72.6, 72.6, 72.9, 73.0, 73.7, 73.7, 73.8, 74.2, 74.2, 74.2, 74.3, 74.3, 74.3, 74.4, 74.6, 74.6, 75.0, 75.1, 77.2, 77.6, 132.3, 132.6, 132.7, 133.1, 133.1, 133.2, 133.2, 133.2, 133.2, 133.7, 134.1, 134.3, 134.4, 134.5, 134.5, 134.6, 134.6, 134.7, 134.7, 134.8, 135.0, 135.2, 135.8, 137.9, 138.1 ], "tpot_quantiles_ms": [ 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.2, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.3, 63.6, 63.6, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8 ] }, { "x": 4.0, "output_throughput": { "median": 61.77, "min": 61.76, "max": 61.8, "n": 3 }, "median_ttft_ms": { "median": 139.88, "min": 138.84, "max": 140.21, "n": 3 }, "p99_ttft_ms": { "median": 141.22, "min": 140.06, "max": 142.01, "n": 3 }, "median_tpot_ms": { "median": 64.11, "min": 64.08, "max": 64.11, "n": 3 }, "p99_tpot_ms": { "median": 64.64, "min": 64.62, "max": 64.65, "n": 3 }, "p99_itl_ms": { "median": 65.64, "min": 65.55, "max": 66.14, "n": 3 }, "median_itl_ms": { "median": 64.08, "min": 64.06, "max": 64.08, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1500.0, "max": 1560.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 69.9, 69.9, 72.9, 73.0, 73.2, 73.4, 74.0, 74.3, 74.6, 74.9, 75.5, 75.6, 83.4, 136.6, 136.7, 136.9, 138.0, 138.0, 138.3, 138.5, 138.8, 138.8, 139.2, 139.3, 139.4, 139.4, 139.5, 139.7, 139.8, 139.8, 139.9, 139.9, 140.0, 140.1, 140.1, 140.1, 140.2, 140.3, 140.4, 140.5, 140.5, 140.6, 140.8, 140.8, 140.8, 141.0, 141.1, 141.1, 141.2, 141.6, 142.1 ], "tpot_quantiles_ms": [ 64.0, 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.5, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.7 ] }, { "x": 8.0, "output_throughput": { "median": 112.9, "min": 112.64, "max": 112.94, "n": 3 }, "median_ttft_ms": { "median": 187.16, "min": 161.37, "max": 189.54, "n": 3 }, "p99_ttft_ms": { "median": 1301.98, "min": 1210.42, "max": 1326.9, "n": 3 }, "median_tpot_ms": { "median": 65.83, "min": 65.79, "max": 66.23, "n": 3 }, "p99_tpot_ms": { "median": 74.43, "min": 73.95, "max": 74.55, "n": 3 }, "p99_itl_ms": { "median": 71.8, "min": 69.95, "max": 72.34, "n": 3 }, "median_itl_ms": { "median": 65.76, "min": 65.7, "max": 65.78, "n": 3 }, "goodput_rps": { "median": 0.61, "min": 0.61, "max": 0.61, "n": 3 }, "request_throughput": { "median": 0.88, "min": 0.88, "max": 0.88, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 74.7, 76.3, 76.6, 83.6, 141.6, 142.8, 143.9, 144.9, 145.0, 145.4, 145.6, 145.7, 146.1, 146.3, 146.5, 146.6, 146.9, 147.1, 147.5, 148.0, 153.4, 153.7, 154.3, 155.0, 173.1, 174.4, 178.0, 203.8, 208.4, 477.3, 478.6, 478.8, 480.4, 481.4, 510.2, 1191.8, 1192.8, 1194.2, 1195.6, 1196.5, 1208.4, 1208.4, 1209.1, 1209.6, 1210.0, 1210.6, 1300.6, 1301.4, 1325.0, 1325.7, 1327.0 ], "tpot_quantiles_ms": [ 65.7, 65.7, 65.7, 65.7, 65.7, 65.7, 65.7, 65.7, 65.7, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 65.8, 66.0, 66.0, 66.0, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.3, 66.3, 66.3, 66.8, 66.8, 71.4, 71.5, 71.5, 71.6, 72.3, 72.5, 73.9, 74.0, 74.9 ] }, { "x": 16.0, "output_throughput": { "median": 201.33, "min": 201.23, "max": 201.81, "n": 3 }, "median_ttft_ms": { "median": 222.41, "min": 220.98, "max": 223.62, "n": 3 }, "p99_ttft_ms": { "median": 2439.42, "min": 2372.95, "max": 2450.09, "n": 3 }, "median_tpot_ms": { "median": 70.14, "min": 70.1, "max": 70.15, "n": 3 }, "p99_tpot_ms": { "median": 87.16, "min": 86.86, "max": 87.23, "n": 3 }, "p99_itl_ms": { "median": 79.96, "min": 78.7, "max": 81.71, "n": 3 }, "median_itl_ms": { "median": 69.57, "min": 69.49, "max": 69.62, "n": 3 }, "goodput_rps": { "median": 0.93, "min": 0.93, "max": 0.94, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.57, "max": 1.58, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1440.0, "max": 1462.5, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 77.3, 80.2, 142.6, 143.9, 144.9, 146.7, 148.1, 148.9, 149.3, 150.1, 150.9, 178.3, 184.2, 212.0, 212.7, 214.0, 215.0, 215.3, 216.5, 218.3, 219.8, 220.5, 221.1, 221.4, 221.9, 222.6, 223.5, 474.3, 476.5, 484.9, 1616.9, 1618.2, 1619.2, 1620.3, 1621.4, 1626.3, 1635.0, 1636.6, 1643.0, 1644.5, 1646.8, 1656.5, 1658.4, 2337.8, 2351.4, 2370.8, 2373.4, 2380.2, 2438.8, 2448.6, 2450.3 ], "tpot_quantiles_ms": [ 69.5, 69.5, 69.5, 69.6, 69.6, 69.6, 69.6, 69.6, 69.6, 69.6, 69.6, 69.6, 69.6, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.8, 69.8, 69.9, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.2, 70.4, 70.4, 70.7, 70.9, 75.3, 75.3, 75.4, 75.4, 75.4, 75.4, 75.5, 75.5, 76.2, 76.2, 76.3, 76.3, 84.4, 84.6, 85.3, 86.8, 87.7 ] }, { "x": 32.0, "output_throughput": { "median": 270.91, "min": 270.52, "max": 272.52, "n": 3 }, "median_ttft_ms": { "median": 2833.4, "min": 2781.94, "max": 2858.53, "n": 3 }, "p99_ttft_ms": { "median": 4805.72, "min": 4770.93, "max": 4817.11, "n": 3 }, "median_tpot_ms": { "median": 96.99, "min": 96.85, "max": 97.03, "n": 3 }, "p99_tpot_ms": { "median": 117.68, "min": 117.51, "max": 118.25, "n": 3 }, "p99_itl_ms": { "median": 102.96, "min": 102.94, "max": 105.37, "n": 3 }, "median_itl_ms": { "median": 81.79, "min": 81.7, "max": 81.81, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.12, "min": 2.11, "max": 2.13, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1320.0, "max": 1327.5, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 106.7, 187.0, 475.7, 486.5, 636.4, 995.3, 1518.0, 1520.6, 1610.9, 1615.8, 1620.1, 1642.2, 1646.9, 1656.6, 1674.6, 1772.9, 1910.8, 2143.1, 2672.4, 2680.7, 2743.5, 2751.6, 2784.9, 2796.4, 2813.3, 2831.5, 2837.8, 2865.1, 2919.3, 3057.2, 3290.4, 3841.8, 3852.3, 3898.4, 3949.7, 3966.0, 3983.5, 4000.7, 4026.0, 4055.6, 4064.4, 4067.0, 4204.0, 4431.6, 4669.3, 4671.7, 4718.0, 4730.1, 4799.7, 4805.5, 4817.5 ], "tpot_quantiles_ms": [ 81.8, 81.9, 82.0, 82.0, 82.0, 82.1, 82.1, 82.4, 83.1, 83.2, 84.7, 87.7, 87.7, 87.8, 88.0, 88.0, 88.3, 88.6, 88.7, 89.6, 91.4, 92.1, 93.7, 96.8, 97.0, 97.0, 97.4, 97.4, 97.4, 97.4, 97.9, 97.9, 98.9, 100.4, 101.2, 102.7, 106.0, 106.2, 106.2, 106.3, 106.8, 106.8, 107.1, 107.1, 108.0, 109.4, 111.7, 115.3, 116.1, 117.4, 118.6 ] }, { "x": 64.0, "output_throughput": { "median": 257.29, "min": 256.73, "max": 258.63, "n": 3 }, "median_ttft_ms": { "median": 18259.53, "min": 18227.21, "max": 18266.8, "n": 3 }, "p99_ttft_ms": { "median": 30973.42, "min": 30730.78, "max": 31540.49, "n": 3 }, "median_tpot_ms": { "median": 107.76, "min": 107.03, "max": 108.42, "n": 3 }, "p99_tpot_ms": { "median": 156.83, "min": 156.77, "max": 161.01, "n": 3 }, "p99_itl_ms": { "median": 112.31, "min": 112.07, "max": 139.15, "n": 3 }, "median_itl_ms": { "median": 84.32, "min": 84.29, "max": 84.53, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.01, "min": 2.01, "max": 2.02, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1215.0, "min": 1170.0, "max": 1230.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.327 }, "ttft_quantiles_ms": [ 189.1, 1235.4, 1679.7, 2651.3, 3505.2, 3968.3, 4647.3, 4681.5, 4704.9, 5117.7, 5312.7, 5409.3, 5426.8, 5633.6, 6310.6, 7591.2, 9756.3, 10825.1, 11155.6, 17113.0, 17231.1, 17261.0, 17269.7, 17408.3, 17468.9, 18228.8, 18268.1, 18412.9, 18429.8, 18617.4, 18875.2, 19400.8, 19417.1, 19589.2, 19594.4, 19812.3, 20282.9, 20329.4, 20575.9, 20764.2, 20879.4, 22620.7, 22950.1, 24298.6, 24883.7, 25613.0, 27012.2, 27229.8, 29632.6, 30307.7, 33043.3 ], "tpot_quantiles_ms": [ 84.2, 84.4, 84.4, 84.5, 84.6, 85.8, 86.5, 89.6, 90.1, 90.2, 90.9, 91.1, 91.4, 93.2, 94.5, 94.8, 96.3, 98.1, 99.2, 99.4, 99.4, 100.0, 100.5, 101.1, 104.6, 107.4, 108.4, 108.6, 108.6, 109.0, 109.8, 110.2, 113.5, 115.2, 116.3, 117.7, 117.8, 118.0, 118.3, 119.3, 119.7, 120.7, 123.9, 131.2, 133.9, 135.3, 137.4, 152.2, 153.9, 156.8, 161.0 ] } ] }, "no-cuda-graph": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 13.81, "min": 13.73, "max": 13.86, "n": 3 }, "median_ttft_ms": { "median": 201.0, "min": 200.11, "max": 204.55, "n": 3 }, "p99_ttft_ms": { "median": 686.89, "min": 682.18, "max": 702.0, "n": 3 }, "median_tpot_ms": { "median": 60.63, "min": 60.62, "max": 60.64, "n": 3 }, "p99_tpot_ms": { "median": 102.51, "min": 102.21, "max": 103.39, "n": 3 }, "p99_itl_ms": { "median": 103.95, "min": 103.73, "max": 104.26, "n": 3 }, "median_itl_ms": { "median": 60.91, "min": 60.87, "max": 60.92, "n": 3 }, "goodput_rps": { "median": 0.1, "min": 0.09, "max": 0.1, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.11, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1290.0, "max": 1320.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.442 }, "ttft_quantiles_ms": [ 90.2, 90.2, 92.1, 92.9, 178.6, 189.4, 189.8, 191.2, 194.6, 195.8, 196.6, 196.7, 198.0, 198.3, 198.5, 199.0, 199.3, 199.3, 199.6, 199.6, 199.7, 199.7, 200.0, 200.2, 200.5, 201.8, 201.8, 202.7, 203.4, 204.3, 204.8, 216.9, 264.9, 314.7, 314.7, 322.8, 485.4, 494.7, 500.1, 500.6, 501.6, 501.8, 503.3, 504.9, 506.0, 506.0, 511.2, 646.6, 711.8, 713.5, 717.9 ], "tpot_quantiles_ms": [ 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.8, 60.8, 60.8, 61.6, 63.7, 63.7, 66.8, 84.8, 85.3, 85.5, 85.5, 85.8, 85.9, 86.1, 86.2, 86.5, 95.2, 97.6, 103.4, 103.4, 103.4, 103.4 ] }, { "x": 2.0, "output_throughput": { "median": 30.82, "min": 30.82, "max": 30.84, "n": 3 }, "median_ttft_ms": { "median": 125.78, "min": 124.62, "max": 127.9, "n": 3 }, "p99_ttft_ms": { "median": 155.54, "min": 153.66, "max": 158.36, "n": 3 }, "median_tpot_ms": { "median": 64.37, "min": 64.33, "max": 64.38, "n": 3 }, "p99_tpot_ms": { "median": 64.68, "min": 64.63, "max": 64.69, "n": 3 }, "p99_itl_ms": { "median": 66.06, "min": 65.02, "max": 67.11, "n": 3 }, "median_itl_ms": { "median": 64.27, "min": 64.24, "max": 64.28, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1515.0, "max": 1530.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.0, 96.0, 96.0, 96.2, 96.6, 96.7, 96.7, 96.7, 96.7, 96.8, 96.9, 97.0, 97.1, 97.4, 97.5, 97.5, 97.7, 97.7, 98.1, 98.1, 98.3, 98.7, 99.2, 99.4, 99.4, 102.6, 152.5, 152.6, 152.9, 152.9, 153.2, 153.3, 153.3, 153.3, 153.3, 153.6, 153.6, 153.7, 153.7, 153.8, 153.8, 153.8, 154.3, 154.4, 154.6, 155.0, 155.2, 155.3, 155.7, 155.7, 158.8 ], "tpot_quantiles_ms": [ 64.0, 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.7, 64.7, 64.7, 64.7, 64.7 ] }, { "x": 4.0, "output_throughput": { "median": 60.86, "min": 60.85, "max": 60.86, "n": 3 }, "median_ttft_ms": { "median": 158.39, "min": 157.88, "max": 158.41, "n": 3 }, "p99_ttft_ms": { "median": 160.54, "min": 160.09, "max": 161.12, "n": 3 }, "median_tpot_ms": { "median": 64.95, "min": 64.94, "max": 64.95, "n": 3 }, "p99_tpot_ms": { "median": 65.42, "min": 65.41, "max": 65.43, "n": 3 }, "p99_itl_ms": { "median": 66.47, "min": 66.05, "max": 67.28, "n": 3 }, "median_itl_ms": { "median": 65.06, "min": 65.03, "max": 65.08, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1560.0, "min": 1515.0, "max": 1560.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 98.3, 98.3, 98.4, 98.8, 98.9, 98.9, 99.0, 99.3, 99.4, 99.8, 100.0, 100.8, 102.0, 156.8, 157.5, 157.6, 157.6, 157.6, 157.7, 157.8, 157.9, 158.0, 158.0, 158.1, 158.2, 158.2, 158.2, 158.3, 158.3, 158.4, 158.6, 158.6, 158.6, 158.7, 158.7, 158.7, 158.8, 158.8, 158.8, 159.1, 159.2, 159.2, 159.2, 159.2, 159.7, 160.0, 160.2, 160.3, 160.6, 160.9, 161.2 ], "tpot_quantiles_ms": [ 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.3, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4 ] }, { "x": 8.0, "output_throughput": { "median": 111.6, "min": 111.48, "max": 111.74, "n": 3 }, "median_ttft_ms": { "median": 187.21, "min": 178.09, "max": 202.94, "n": 3 }, "p99_ttft_ms": { "median": 1266.97, "min": 1226.64, "max": 1299.6, "n": 3 }, "median_tpot_ms": { "median": 66.64, "min": 66.61, "max": 66.8, "n": 3 }, "p99_tpot_ms": { "median": 74.92, "min": 74.52, "max": 75.29, "n": 3 }, "p99_itl_ms": { "median": 68.14, "min": 67.8, "max": 69.28, "n": 3 }, "median_itl_ms": { "median": 66.66, "min": 66.65, "max": 66.74, "n": 3 }, "goodput_rps": { "median": 0.63, "min": 0.49, "max": 0.63, "n": 3 }, "request_throughput": { "median": 0.87, "min": 0.87, "max": 0.87, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1500.0, "max": 1530.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 100.9, 101.0, 101.9, 117.3, 159.8, 160.5, 160.8, 161.5, 161.8, 162.4, 162.5, 162.7, 163.0, 163.1, 163.4, 163.5, 163.8, 164.0, 164.4, 164.5, 165.1, 165.1, 177.6, 178.2, 178.8, 191.1, 198.9, 210.7, 218.4, 932.3, 933.6, 934.5, 937.9, 939.2, 1073.9, 1074.8, 1076.2, 1117.6, 1118.6, 1120.0, 1198.1, 1198.8, 1200.1, 1201.4, 1226.2, 1238.2, 1239.3, 1295.9, 1297.0, 1298.4, 1299.9 ], "tpot_quantiles_ms": [ 66.4, 66.4, 66.4, 66.4, 66.4, 66.4, 66.4, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.8, 66.8, 66.8, 66.8, 66.9, 67.0, 67.0, 67.0, 67.0, 67.1, 67.8, 67.9, 67.9, 67.9, 67.9, 67.9, 67.9, 68.8, 68.8, 69.0, 69.0, 69.0, 74.5, 74.8, 75.5 ] }, { "x": 16.0, "output_throughput": { "median": 200.27, "min": 199.35, "max": 200.45, "n": 3 }, "median_ttft_ms": { "median": 234.3, "min": 234.12, "max": 235.23, "n": 3 }, "p99_ttft_ms": { "median": 2397.54, "min": 2382.33, "max": 2446.67, "n": 3 }, "median_tpot_ms": { "median": 70.77, "min": 70.49, "max": 70.91, "n": 3 }, "p99_tpot_ms": { "median": 87.43, "min": 87.25, "max": 87.95, "n": 3 }, "p99_itl_ms": { "median": 71.92, "min": 71.68, "max": 72.25, "n": 3 }, "median_itl_ms": { "median": 70.17, "min": 70.13, "max": 70.32, "n": 3 }, "goodput_rps": { "median": 0.83, "min": 0.83, "max": 1.03, "n": 3 }, "request_throughput": { "median": 1.56, "min": 1.56, "max": 1.57, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 101.5, 106.8, 162.3, 164.9, 167.1, 168.5, 170.1, 172.2, 172.6, 173.4, 174.1, 174.6, 204.3, 219.8, 221.8, 223.0, 224.4, 225.9, 227.0, 228.9, 232.5, 232.8, 233.2, 233.5, 234.1, 234.5, 235.1, 633.0, 944.7, 1052.7, 1056.2, 1059.2, 1070.0, 1072.9, 1233.9, 1237.6, 1785.7, 1787.6, 2105.3, 2106.9, 2218.0, 2220.8, 2221.8, 2223.8, 2234.4, 2236.2, 2340.4, 2381.5, 2401.5, 2416.0, 2496.6 ], "tpot_quantiles_ms": [ 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.1, 70.1, 70.1, 70.2, 70.2, 70.4, 70.4, 70.4, 70.4, 70.5, 70.5, 70.5, 70.5, 70.5, 70.5, 70.8, 70.8, 71.0, 71.0, 71.0, 71.5, 71.5, 71.5, 71.5, 71.5, 71.5, 72.5, 72.5, 74.6, 74.6, 74.6, 80.3, 80.6, 80.6, 80.6, 80.6, 80.8, 81.6, 83.7, 87.4, 88.5 ] }, { "x": 32.0, "output_throughput": { "median": 270.8, "min": 269.47, "max": 272.06, "n": 3 }, "median_ttft_ms": { "median": 3240.62, "min": 2938.16, "max": 3290.33, "n": 3 }, "p99_ttft_ms": { "median": 4795.72, "min": 4759.81, "max": 4824.62, "n": 3 }, "median_tpot_ms": { "median": 94.11, "min": 93.82, "max": 94.12, "n": 3 }, "p99_tpot_ms": { "median": 118.05, "min": 117.47, "max": 118.06, "n": 3 }, "p99_itl_ms": { "median": 89.41, "min": 88.98, "max": 96.23, "n": 3 }, "median_itl_ms": { "median": 82.13, "min": 82.05, "max": 82.2, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.12, "min": 2.11, "max": 2.13, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1335.0, "min": 1335.0, "max": 1342.5, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 116.3, 212.4, 658.6, 935.1, 939.0, 1056.0, 1058.9, 1066.7, 1097.7, 1099.5, 1375.3, 1784.1, 2093.3, 2105.8, 2120.4, 2126.2, 2202.9, 2205.7, 2209.1, 2241.2, 2250.5, 2258.3, 2261.7, 2547.7, 2938.7, 3240.4, 3289.3, 3291.8, 3338.4, 3342.8, 3345.3, 3400.2, 3408.2, 3414.4, 3430.2, 3746.2, 4094.7, 4258.8, 4429.9, 4432.2, 4468.0, 4493.3, 4495.9, 4497.7, 4541.2, 4556.7, 4603.3, 4613.2, 4673.0, 4759.6, 4842.2 ], "tpot_quantiles_ms": [ 82.0, 82.1, 82.2, 82.3, 83.3, 83.3, 83.5, 83.5, 83.5, 83.5, 83.8, 84.1, 84.5, 84.6, 84.8, 89.6, 90.1, 92.4, 92.6, 92.6, 92.8, 92.9, 92.9, 93.1, 93.8, 94.1, 94.1, 99.1, 99.1, 101.4, 101.5, 101.5, 101.7, 101.9, 102.1, 102.1, 102.2, 103.1, 103.4, 103.6, 108.2, 108.3, 110.5, 110.8, 111.0, 111.2, 111.4, 112.4, 113.1, 117.5, 118.8 ] }, { "x": 64.0, "output_throughput": { "median": 256.41, "min": 254.76, "max": 258.04, "n": 3 }, "median_ttft_ms": { "median": 18342.99, "min": 18329.17, "max": 18389.8, "n": 3 }, "p99_ttft_ms": { "median": 31627.55, "min": 31298.86, "max": 32354.52, "n": 3 }, "median_tpot_ms": { "median": 106.7, "min": 106.34, "max": 108.04, "n": 3 }, "p99_tpot_ms": { "median": 155.2, "min": 153.25, "max": 162.8, "n": 3 }, "p99_itl_ms": { "median": 97.69, "min": 94.1, "max": 108.48, "n": 3 }, "median_itl_ms": { "median": 84.37, "min": 84.13, "max": 84.48, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.0, "min": 1.99, "max": 2.02, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1230.0, "min": 1200.0, "max": 1230.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.349 }, "ttft_quantiles_ms": [ 194.7, 1692.1, 1840.4, 2843.0, 3993.2, 4150.7, 4718.0, 4752.9, 4837.4, 5166.4, 5380.3, 5445.1, 5491.2, 5550.0, 6390.2, 7506.4, 10280.5, 11000.0, 11427.3, 17172.9, 17236.6, 17299.1, 17358.6, 17410.5, 17505.7, 18330.8, 18391.0, 18473.2, 18534.7, 18579.3, 19073.2, 19500.7, 19542.8, 19655.6, 19712.9, 19757.8, 20545.2, 20660.4, 20785.6, 20843.6, 20897.0, 22514.1, 23156.0, 24749.9, 25642.8, 25798.8, 27421.7, 27988.1, 30084.9, 31146.8, 33829.5 ], "tpot_quantiles_ms": [ 84.0, 84.1, 84.2, 84.3, 84.4, 84.9, 85.3, 89.5, 89.8, 89.9, 90.0, 90.0, 90.3, 93.2, 94.1, 94.6, 94.8, 94.9, 98.9, 99.2, 99.3, 99.4, 99.6, 103.1, 103.7, 106.7, 108.1, 108.3, 108.6, 108.6, 108.9, 110.6, 112.2, 112.8, 114.8, 117.2, 117.5, 117.9, 118.0, 118.3, 120.2, 121.6, 124.2, 131.7, 132.6, 134.4, 137.1, 150.7, 154.2, 155.2, 162.8 ] } ] }, "dp2-tokens": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 15.15, "min": 14.83, "max": 15.77, "n": 3 }, "median_ttft_ms": { "median": 182.95, "min": 181.54, "max": 184.58, "n": 3 }, "p99_ttft_ms": { "median": 476.99, "min": 471.47, "max": 506.24, "n": 3 }, "median_tpot_ms": { "median": 60.14, "min": 60.1, "max": 60.14, "n": 3 }, "p99_tpot_ms": { "median": 84.28, "min": 82.54, "max": 91.03, "n": 3 }, "p99_itl_ms": { "median": 90.67, "min": 85.69, "max": 94.12, "n": 3 }, "median_itl_ms": { "median": 60.16, "min": 60.11, "max": 60.16, "n": 3 }, "goodput_rps": { "median": 0.12, "min": 0.12, "max": 0.12, "n": 3 }, "request_throughput": { "median": 0.12, "min": 0.12, "max": 0.12, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1965.0, "min": 1410.0, "max": 2025.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.158 }, "ttft_quantiles_ms": [ 72.1, 72.1, 72.4, 72.9, 168.4, 169.6, 171.9, 173.1, 175.1, 175.7, 176.6, 178.1, 178.1, 179.0, 179.4, 180.3, 180.4, 180.4, 180.5, 180.9, 180.9, 181.2, 181.5, 181.6, 181.7, 182.2, 183.7, 183.7, 184.3, 184.5, 184.9, 185.0, 185.2, 185.7, 185.7, 186.8, 187.5, 188.9, 189.9, 200.7, 205.9, 287.4, 380.1, 412.5, 415.8, 453.4, 467.6, 468.8, 471.9, 494.1, 513.1 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.2, 60.4, 60.4, 60.4, 62.8, 67.4, 67.4, 81.1, 81.5, 82.7, 87.3, 90.6, 91.0, 91.0 ] }, { "x": 2.0, "output_throughput": { "median": 21.12, "min": 19.74, "max": 22.48, "n": 3 }, "median_ttft_ms": { "median": 94.02, "min": 78.8, "max": 97.75, "n": 3 }, "p99_ttft_ms": { "median": 507.38, "min": 503.74, "max": 533.7, "n": 3 }, "median_tpot_ms": { "median": 94.89, "min": 84.66, "max": 99.58, "n": 3 }, "p99_tpot_ms": { "median": 147.42, "min": 129.51, "max": 151.15, "n": 3 }, "p99_itl_ms": { "median": 217.47, "min": 214.4, "max": 218.0, "n": 3 }, "median_itl_ms": { "median": 61.02, "min": 60.77, "max": 61.62, "n": 3 }, "goodput_rps": { "median": 0.09, "min": 0.09, "max": 0.13, "n": 3 }, "request_throughput": { "median": 0.16, "min": 0.15, "max": 0.18, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 870.0, "min": 780.0, "max": 1200.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.445 }, "ttft_quantiles_ms": [ 71.9, 71.9, 72.1, 72.7, 72.8, 72.8, 72.8, 72.9, 72.9, 72.9, 73.0, 73.1, 73.2, 73.2, 73.3, 73.4, 73.5, 73.5, 74.0, 74.0, 74.9, 78.6, 79.0, 87.5, 87.9, 90.7, 92.8, 93.1, 95.2, 102.4, 109.9, 173.9, 178.6, 185.0, 185.0, 238.0, 261.2, 357.7, 361.0, 397.8, 408.4, 412.0, 413.7, 416.5, 418.7, 419.7, 426.4, 486.1, 511.1, 517.4, 553.8 ], "tpot_quantiles_ms": [ 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.2, 60.2, 60.2, 60.2, 60.4, 60.4, 60.5, 60.6, 60.6, 62.6, 76.5, 82.1, 83.6, 84.0, 85.3, 90.3, 96.7, 98.6, 99.0, 99.3, 99.5, 99.7, 99.9, 100.3, 100.4, 100.4, 101.1, 101.2, 102.0, 102.2, 102.2, 103.3, 103.4, 103.4, 103.6, 103.8, 107.0, 114.2, 133.5, 137.7, 153.3, 153.5 ] }, { "x": 4.0, "output_throughput": { "median": 48.54, "min": 48.12, "max": 48.55, "n": 3 }, "median_ttft_ms": { "median": 134.04, "min": 134.03, "max": 136.33, "n": 3 }, "p99_ttft_ms": { "median": 647.95, "min": 629.04, "max": 710.26, "n": 3 }, "median_tpot_ms": { "median": 63.85, "min": 63.83, "max": 63.85, "n": 3 }, "p99_tpot_ms": { "median": 109.61, "min": 109.4, "max": 111.73, "n": 3 }, "p99_itl_ms": { "median": 215.37, "min": 215.16, "max": 220.0, "n": 3 }, "median_itl_ms": { "median": 63.64, "min": 63.63, "max": 63.64, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.23, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.38, "min": 0.38, "max": 0.38, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1455.0, "max": 1485.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.015 }, "ttft_quantiles_ms": [ 97.7, 97.7, 98.3, 98.5, 99.0, 99.0, 99.1, 99.2, 99.6, 100.3, 100.5, 100.7, 100.8, 101.0, 101.8, 102.9, 132.6, 132.6, 133.1, 133.5, 133.6, 133.7, 133.8, 133.8, 134.3, 134.3, 134.4, 135.6, 136.3, 136.3, 136.7, 137.6, 192.3, 202.7, 202.7, 203.0, 208.4, 224.3, 225.7, 231.4, 248.5, 306.6, 570.6, 571.5, 598.5, 600.2, 603.2, 604.5, 633.6, 655.6, 729.7 ], "tpot_quantiles_ms": [ 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.9, 63.9, 63.9, 63.9, 63.9, 104.2, 104.9, 104.9, 105.3, 105.9, 106.0, 106.2, 106.5, 106.5, 107.4, 107.8, 108.1, 109.1, 109.2, 109.2, 109.4, 109.6, 109.7, 112.1 ] }, { "x": 8.0, "output_throughput": { "median": 92.09, "min": 89.8, "max": 92.59, "n": 3 }, "median_ttft_ms": { "median": 519.58, "min": 500.74, "max": 556.28, "n": 3 }, "p99_ttft_ms": { "median": 1296.57, "min": 1275.87, "max": 1438.19, "n": 3 }, "median_tpot_ms": { "median": 65.81, "min": 65.74, "max": 65.97, "n": 3 }, "p99_tpot_ms": { "median": 115.99, "min": 115.13, "max": 120.31, "n": 3 }, "p99_itl_ms": { "median": 213.37, "min": 208.9, "max": 230.08, "n": 3 }, "median_itl_ms": { "median": 64.52, "min": 64.51, "max": 64.55, "n": 3 }, "goodput_rps": { "median": 0.45, "min": 0.44, "max": 0.45, "n": 3 }, "request_throughput": { "median": 0.72, "min": 0.7, "max": 0.72, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1455.0, "max": 1470.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.022 }, "ttft_quantiles_ms": [ 87.3, 88.2, 90.2, 91.1, 138.1, 138.7, 139.2, 139.9, 140.0, 140.7, 141.1, 141.5, 166.4, 191.4, 194.2, 195.0, 197.0, 197.6, 224.9, 349.7, 389.4, 424.8, 439.5, 496.7, 497.2, 504.9, 510.6, 529.3, 615.4, 616.4, 618.4, 618.9, 622.1, 662.4, 674.7, 688.2, 689.4, 822.4, 913.8, 939.4, 940.9, 986.7, 1010.2, 1011.7, 1173.5, 1266.0, 1266.7, 1296.0, 1296.7, 1329.0, 1486.9 ], "tpot_quantiles_ms": [ 64.3, 64.3, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.5, 64.6, 64.6, 64.6, 64.8, 64.8, 64.8, 65.7, 65.7, 65.8, 65.8, 66.0, 67.7, 67.8, 68.2, 68.3, 68.4, 106.1, 106.1, 106.1, 106.8, 106.8, 107.4, 107.5, 107.5, 108.4, 108.6, 108.9, 110.6, 111.0, 111.0, 114.0, 114.3, 115.6, 117.2, 121.7 ] }, { "x": 16.0, "output_throughput": { "median": 73.67, "min": 73.26, "max": 75.95, "n": 3 }, "median_ttft_ms": { "median": 538.86, "min": 483.82, "max": 702.0, "n": 3 }, "p99_ttft_ms": { "median": 43319.22, "min": 42508.89, "max": 45545.76, "n": 3 }, "median_tpot_ms": { "median": 70.75, "min": 70.62, "max": 70.97, "n": 3 }, "p99_tpot_ms": { "median": 165.49, "min": 160.02, "max": 176.34, "n": 3 }, "p99_itl_ms": { "median": 246.4, "min": 246.32, "max": 247.46, "n": 3 }, "median_itl_ms": { "median": 65.91, "min": 65.83, "max": 65.91, "n": 3 }, "goodput_rps": { "median": 0.38, "min": 0.37, "max": 0.39, "n": 3 }, "request_throughput": { "median": 0.58, "min": 0.57, "max": 0.59, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1395.0, "max": 1410.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.265 }, "ttft_quantiles_ms": [ 87.7, 90.0, 138.0, 138.8, 139.4, 140.2, 141.0, 142.0, 156.7, 158.3, 196.7, 205.6, 206.1, 207.5, 291.4, 304.4, 307.8, 411.8, 435.1, 438.7, 439.6, 443.4, 444.3, 502.9, 523.3, 573.7, 577.8, 692.0, 701.7, 703.1, 713.1, 724.4, 727.5, 742.7, 744.7, 1141.3, 12852.0, 13175.6, 15665.1, 15720.5, 15755.6, 15904.7, 15963.7, 20481.1, 27729.1, 29205.8, 29707.9, 30122.5, 38802.7, 42776.7, 50259.7 ], "tpot_quantiles_ms": [ 64.3, 64.4, 66.0, 66.0, 66.0, 66.4, 66.4, 66.4, 66.4, 66.5, 66.9, 67.3, 67.4, 68.4, 68.8, 68.8, 68.8, 68.8, 68.8, 68.9, 69.2, 69.2, 69.7, 69.7, 70.6, 70.7, 70.8, 71.0, 71.3, 72.2, 72.6, 73.4, 76.4, 106.0, 106.3, 106.5, 106.8, 107.3, 108.9, 109.9, 110.5, 110.6, 114.1, 127.6, 128.7, 129.7, 155.1, 159.2, 160.1, 165.5, 176.3 ] }, { "x": 32.0, "output_throughput": { "median": 88.33, "min": 88.19, "max": 116.88, "n": 3 }, "median_ttft_ms": { "median": 1018.29, "min": 947.27, "max": 1813.13, "n": 3 }, "p99_ttft_ms": { "median": 110008.42, "min": 70787.98, "max": 113526.13, "n": 3 }, "median_tpot_ms": { "median": 81.21, "min": 80.24, "max": 81.77, "n": 3 }, "p99_tpot_ms": { "median": 178.29, "min": 177.92, "max": 185.64, "n": 3 }, "p99_itl_ms": { "median": 255.08, "min": 247.83, "max": 380.95, "n": 3 }, "median_itl_ms": { "median": 74.47, "min": 69.32, "max": 76.29, "n": 3 }, "goodput_rps": { "median": 0.32, "min": 0.14, "max": 0.35, "n": 3 }, "request_throughput": { "median": 0.69, "min": 0.69, "max": 0.91, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1350.0, "max": 1395.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.359 }, "ttft_quantiles_ms": [ 94.2, 192.5, 309.8, 390.5, 441.9, 486.7, 502.1, 585.5, 618.7, 635.5, 712.4, 715.4, 717.2, 723.0, 730.4, 740.4, 851.2, 867.7, 911.3, 929.4, 1020.3, 1038.8, 1041.7, 1114.9, 1156.2, 1161.7, 1217.1, 1220.5, 1559.1, 1561.8, 1566.0, 1693.1, 1735.6, 1811.2, 1824.4, 1829.1, 2276.6, 2290.4, 2812.9, 15700.2, 15970.7, 30654.8, 30936.2, 45855.3, 46830.7, 52269.3, 65428.0, 71540.6, 89646.8, 97796.0, 119201.2 ], "tpot_quantiles_ms": [ 64.7, 68.0, 69.8, 71.3, 72.0, 73.3, 74.3, 74.5, 76.0, 76.3, 76.3, 76.3, 76.4, 76.4, 76.5, 76.7, 77.7, 77.7, 78.5, 78.8, 79.1, 79.3, 79.7, 80.2, 81.0, 81.2, 81.4, 81.6, 81.8, 82.8, 84.2, 85.5, 85.9, 86.6, 87.9, 88.5, 90.7, 95.0, 106.9, 107.3, 108.8, 110.8, 113.7, 132.8, 148.0, 156.3, 165.2, 167.1, 171.4, 177.9, 185.6 ] }, { "x": 64.0, "output_throughput": { "median": 114.69, "min": 112.66, "max": 114.87, "n": 3 }, "median_ttft_ms": { "median": 3976.51, "min": 3916.0, "max": 4118.96, "n": 3 }, "p99_ttft_ms": { "median": 171192.13, "min": 170391.29, "max": 178777.25, "n": 3 }, "median_tpot_ms": { "median": 105.32, "min": 100.74, "max": 105.72, "n": 3 }, "p99_tpot_ms": { "median": 182.33, "min": 177.01, "max": 184.34, "n": 3 }, "p99_itl_ms": { "median": 252.14, "min": 248.53, "max": 253.08, "n": 3 }, "median_itl_ms": { "median": 82.6, "min": 82.28, "max": 82.84, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.9, "min": 0.88, "max": 0.9, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1020.0, "min": 847.5, "max": 1320.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.432 }, "ttft_quantiles_ms": [ 184.9, 213.3, 429.7, 615.8, 835.5, 1272.9, 1551.5, 1630.6, 1672.5, 1697.5, 1716.2, 1778.2, 1789.9, 1862.0, 2643.8, 2728.4, 2796.5, 2845.5, 2863.9, 2899.0, 2961.3, 2969.9, 3026.4, 3593.3, 3824.0, 3937.1, 4021.3, 4061.1, 4153.5, 4168.8, 4320.5, 4635.3, 4718.1, 4768.3, 4831.1, 4883.3, 5021.5, 5402.1, 5680.8, 6564.1, 8006.9, 8374.1, 15883.8, 45959.9, 67382.7, 90054.2, 121718.7, 145424.1, 156634.6, 169475.2, 180684.6 ], "tpot_quantiles_ms": [ 63.3, 81.7, 81.8, 82.4, 82.5, 83.3, 83.6, 84.1, 87.0, 87.4, 88.0, 88.2, 88.6, 89.7, 91.3, 93.0, 95.9, 96.5, 96.8, 97.3, 97.5, 97.9, 99.5, 99.9, 103.1, 105.1, 105.7, 106.5, 106.8, 107.0, 107.2, 107.8, 108.9, 110.6, 111.2, 112.5, 114.4, 114.8, 116.1, 118.2, 123.0, 125.4, 127.7, 142.0, 148.6, 156.2, 162.2, 167.1, 174.0, 177.1, 202.0 ] } ] }, "graphs-on-b": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 13.93, "min": 13.71, "max": 15.2, "n": 3 }, "median_ttft_ms": { "median": 184.82, "min": 181.57, "max": 188.19, "n": 3 }, "p99_ttft_ms": { "median": 553.05, "min": 435.27, "max": 660.94, "n": 3 }, "median_tpot_ms": { "median": 60.15, "min": 60.15, "max": 60.16, "n": 3 }, "p99_tpot_ms": { "median": 102.35, "min": 77.93, "max": 103.35, "n": 3 }, "p99_itl_ms": { "median": 103.76, "min": 103.3, "max": 103.8, "n": 3 }, "median_itl_ms": { "median": 60.21, "min": 60.19, "max": 60.22, "n": 3 }, "goodput_rps": { "median": 0.1, "min": 0.08, "max": 0.12, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.12, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1290.0, "max": 1305.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.442 }, "ttft_quantiles_ms": [ 71.4, 71.4, 71.4, 72.8, 166.1, 167.3, 172.9, 173.7, 174.5, 174.5, 175.0, 178.2, 178.6, 178.7, 179.4, 180.4, 180.7, 180.7, 180.9, 181.1, 181.5, 181.6, 183.4, 183.4, 183.8, 186.2, 187.9, 188.1, 188.3, 189.5, 189.7, 190.0, 209.9, 299.7, 299.7, 315.0, 357.4, 372.6, 379.9, 418.4, 429.6, 436.3, 473.3, 479.3, 483.9, 489.6, 505.1, 529.2, 551.6, 553.3, 691.2 ], "tpot_quantiles_ms": [ 60.0, 60.0, 60.0, 60.0, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.1, 60.2, 60.2, 60.2, 60.2, 60.3, 60.3, 63.3, 63.3, 63.3, 63.4, 71.9, 72.2, 72.7, 72.9, 78.8, 84.8, 85.3, 85.9, 96.6, 102.7, 103.2, 103.3, 103.4, 103.4 ] }, { "x": 2.0, "output_throughput": { "median": 31.2, "min": 31.2, "max": 31.2, "n": 3 }, "median_ttft_ms": { "median": 115.93, "min": 115.21, "max": 116.12, "n": 3 }, "p99_ttft_ms": { "median": 133.67, "min": 132.9, "max": 134.38, "n": 3 }, "median_tpot_ms": { "median": 63.65, "min": 63.64, "max": 63.66, "n": 3 }, "p99_tpot_ms": { "median": 63.86, "min": 63.86, "max": 63.87, "n": 3 }, "p99_itl_ms": { "median": 64.97, "min": 64.81, "max": 64.98, "n": 3 }, "median_itl_ms": { "median": 63.51, "min": 63.5, "max": 63.52, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1500.0, "max": 1515.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.6, 95.6, 97.1, 97.3, 97.3, 97.5, 97.6, 97.6, 97.6, 97.7, 97.7, 97.8, 97.8, 97.9, 97.9, 98.0, 98.0, 98.0, 98.4, 98.4, 98.4, 98.4, 99.2, 99.4, 99.8, 101.1, 130.8, 131.8, 132.0, 132.2, 132.3, 132.4, 132.4, 132.4, 132.4, 132.5, 132.5, 132.6, 132.6, 132.8, 132.8, 132.8, 132.8, 132.9, 133.0, 133.1, 133.3, 133.6, 133.7, 134.3, 134.4 ], "tpot_quantiles_ms": [ 63.4, 63.4, 63.4, 63.4, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.5, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.6, 63.7, 63.7, 63.7, 63.7, 63.7, 63.7, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.8, 63.9, 63.9, 63.9, 63.9 ] }, { "x": 4.0, "output_throughput": { "median": 61.56, "min": 61.54, "max": 61.57, "n": 3 }, "median_ttft_ms": { "median": 137.24, "min": 136.33, "max": 137.34, "n": 3 }, "p99_ttft_ms": { "median": 138.14, "min": 137.88, "max": 138.31, "n": 3 }, "median_tpot_ms": { "median": 64.36, "min": 64.34, "max": 64.37, "n": 3 }, "p99_tpot_ms": { "median": 64.78, "min": 64.76, "max": 64.8, "n": 3 }, "p99_itl_ms": { "median": 66.24, "min": 65.7, "max": 66.42, "n": 3 }, "median_itl_ms": { "median": 64.33, "min": 64.31, "max": 64.34, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1500.0, "max": 1530.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 84.7, 84.7, 86.1, 86.4, 86.6, 86.7, 86.7, 86.7, 86.7, 86.7, 86.8, 87.2, 87.2, 135.1, 135.7, 136.0, 136.1, 136.1, 136.3, 136.6, 136.6, 136.6, 136.9, 136.9, 137.0, 137.1, 137.1, 137.2, 137.2, 137.2, 137.2, 137.4, 137.4, 137.4, 137.4, 137.5, 137.5, 137.6, 137.6, 137.6, 137.7, 137.8, 137.8, 137.8, 137.9, 138.0, 138.1, 138.1, 138.1, 138.1, 138.4 ], "tpot_quantiles_ms": [ 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.4, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.8 ] }, { "x": 8.0, "output_throughput": { "median": 112.6, "min": 112.49, "max": 112.93, "n": 3 }, "median_ttft_ms": { "median": 165.24, "min": 162.34, "max": 169.39, "n": 3 }, "p99_ttft_ms": { "median": 1312.7, "min": 1210.02, "max": 1334.24, "n": 3 }, "median_tpot_ms": { "median": 66.01, "min": 66.0, "max": 66.08, "n": 3 }, "p99_tpot_ms": { "median": 74.65, "min": 74.07, "max": 74.71, "n": 3 }, "p99_itl_ms": { "median": 69.63, "min": 68.61, "max": 69.9, "n": 3 }, "median_itl_ms": { "median": 65.99, "min": 65.93, "max": 66.04, "n": 3 }, "goodput_rps": { "median": 0.63, "min": 0.58, "max": 0.66, "n": 3 }, "request_throughput": { "median": 0.88, "min": 0.88, "max": 0.88, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1485.0, "max": 1515.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 87.1, 87.8, 89.4, 100.1, 136.9, 137.6, 137.9, 138.5, 139.0, 139.2, 139.8, 140.0, 140.6, 141.0, 141.4, 141.9, 142.1, 142.3, 142.6, 142.9, 143.2, 143.6, 151.8, 152.4, 153.0, 153.7, 185.1, 188.1, 213.8, 494.2, 517.7, 518.3, 618.1, 632.3, 634.2, 638.8, 1171.3, 1172.2, 1198.2, 1199.1, 1208.0, 1208.5, 1209.6, 1225.5, 1226.7, 1228.0, 1311.5, 1312.2, 1332.0, 1333.3, 1334.4 ], "tpot_quantiles_ms": [ 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.0, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.3, 66.3, 66.3, 66.4, 66.4, 70.4, 70.4, 70.5, 70.6, 71.6, 71.6, 72.4, 72.8, 74.0, 74.2, 75.1 ] }, { "x": 16.0, "output_throughput": { "median": 201.21, "min": 200.88, "max": 201.59, "n": 3 }, "median_ttft_ms": { "median": 213.07, "min": 212.1, "max": 216.55, "n": 3 }, "p99_ttft_ms": { "median": 2460.7, "min": 2387.16, "max": 2460.94, "n": 3 }, "median_tpot_ms": { "median": 70.08, "min": 70.03, "max": 70.37, "n": 3 }, "p99_tpot_ms": { "median": 87.23, "min": 86.93, "max": 87.28, "n": 3 }, "p99_itl_ms": { "median": 80.38, "min": 79.26, "max": 80.4, "n": 3 }, "median_itl_ms": { "median": 69.65, "min": 69.63, "max": 69.76, "n": 3 }, "goodput_rps": { "median": 0.98, "min": 0.96, "max": 1.01, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.57, "max": 1.57, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1440.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 90.5, 94.9, 150.3, 151.9, 153.7, 155.2, 156.2, 159.7, 160.3, 160.9, 161.6, 162.4, 163.5, 192.9, 200.6, 202.4, 203.2, 203.8, 205.1, 207.0, 208.7, 211.3, 211.9, 212.4, 213.5, 215.0, 216.4, 485.0, 620.6, 622.4, 625.7, 771.4, 1650.8, 1653.8, 1786.2, 1787.8, 1793.5, 1794.2, 1795.0, 1795.6, 1810.8, 1812.6, 1967.5, 1969.5, 2347.8, 2360.0, 2371.1, 2385.8, 2458.6, 2460.4, 2461.3 ], "tpot_quantiles_ms": [ 69.6, 69.6, 69.6, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.7, 69.8, 69.8, 69.8, 69.8, 70.0, 70.0, 70.0, 70.0, 70.0, 70.0, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.4, 70.4, 70.6, 73.8, 73.9, 74.2, 74.2, 74.2, 74.3, 74.3, 74.3, 74.3, 74.3, 74.4, 76.2, 83.3, 83.3, 83.5, 83.5, 83.7, 86.9, 87.9 ] }, { "x": 32.0, "output_throughput": { "median": 270.61, "min": 270.51, "max": 272.25, "n": 3 }, "median_ttft_ms": { "median": 2928.32, "min": 2880.98, "max": 2996.18, "n": 3 }, "p99_ttft_ms": { "median": 4865.91, "min": 4783.1, "max": 4896.06, "n": 3 }, "median_tpot_ms": { "median": 95.65, "min": 95.23, "max": 96.79, "n": 3 }, "p99_tpot_ms": { "median": 117.79, "min": 117.76, "max": 117.86, "n": 3 }, "p99_itl_ms": { "median": 102.52, "min": 100.81, "max": 102.83, "n": 3 }, "median_itl_ms": { "median": 81.51, "min": 81.51, "max": 81.58, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.11, "min": 2.11, "max": 2.13, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1335.0, "min": 1320.0, "max": 1335.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 106.4, 198.1, 496.2, 506.2, 621.3, 793.1, 1092.4, 1648.9, 1666.0, 1683.9, 1686.5, 1689.1, 1696.1, 1775.9, 1788.9, 1805.1, 1807.7, 1956.3, 2262.1, 2799.9, 2847.9, 2859.4, 2866.4, 2874.7, 2897.0, 2928.1, 2953.0, 2955.1, 2998.5, 3004.4, 3120.8, 3430.0, 3964.1, 4036.4, 4038.5, 4055.3, 4064.4, 4091.8, 4114.5, 4123.3, 4189.4, 4195.6, 4198.7, 4280.3, 4511.6, 4649.4, 4764.1, 4774.6, 4783.5, 4865.7, 4896.4 ], "tpot_quantiles_ms": [ 81.6, 81.6, 81.7, 81.8, 81.8, 81.8, 81.9, 81.9, 82.8, 84.6, 86.1, 86.2, 86.3, 87.2, 87.4, 87.5, 88.3, 88.3, 88.4, 90.1, 92.0, 92.0, 93.8, 95.2, 95.6, 95.8, 96.7, 96.8, 96.8, 97.4, 97.6, 97.7, 99.2, 101.1, 102.9, 104.3, 105.0, 105.1, 106.1, 106.1, 106.3, 106.5, 107.0, 107.1, 108.4, 110.2, 113.5, 114.5, 115.7, 117.4, 118.9 ] }, { "x": 64.0, "output_throughput": { "median": 264.1, "min": 257.27, "max": 265.72, "n": 3 }, "median_ttft_ms": { "median": 18323.66, "min": 18286.81, "max": 18399.78, "n": 3 }, "p99_ttft_ms": { "median": 29689.99, "min": 29285.67, "max": 31619.78, "n": 3 }, "median_tpot_ms": { "median": 103.39, "min": 103.05, "max": 104.72, "n": 3 }, "p99_tpot_ms": { "median": 151.24, "min": 150.26, "max": 157.31, "n": 3 }, "p99_itl_ms": { "median": 109.2, "min": 108.7, "max": 111.56, "n": 3 }, "median_itl_ms": { "median": 83.78, "min": 83.75, "max": 83.92, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.06, "min": 2.01, "max": 2.08, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1245.0, "min": 1215.0, "max": 1245.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.347 }, "ttft_quantiles_ms": [ 189.6, 831.7, 1977.9, 3141.7, 3147.1, 4306.1, 4675.0, 4759.0, 4773.3, 5294.8, 5391.0, 5463.9, 5501.0, 5578.8, 6218.1, 7310.1, 9375.5, 10340.9, 10646.7, 17160.5, 17225.6, 17295.9, 17307.1, 17340.9, 17458.0, 18288.3, 18401.1, 18457.8, 18484.5, 18568.6, 18679.1, 18888.3, 19479.0, 19571.3, 19627.1, 19677.0, 19785.5, 20050.3, 20483.2, 20800.8, 20877.4, 21030.6, 22430.8, 23165.0, 23324.8, 25468.2, 25587.4, 25959.2, 28388.3, 30618.1, 32838.7 ], "tpot_quantiles_ms": [ 83.7, 83.8, 83.9, 83.9, 84.0, 84.1, 84.3, 88.1, 89.6, 89.6, 90.2, 90.5, 90.6, 91.7, 92.1, 92.5, 93.7, 95.5, 98.9, 99.0, 99.4, 99.8, 100.7, 100.8, 103.1, 103.4, 107.9, 108.0, 108.3, 108.4, 108.6, 109.3, 109.9, 110.8, 112.9, 113.9, 117.1, 117.5, 117.7, 117.8, 118.0, 118.7, 123.5, 123.6, 129.1, 130.5, 130.6, 142.9, 150.2, 152.4, 157.3 ] } ] }, "graphs-off-b": { "kv_pool_tokens": 24094, "points": [ { "x": 1.0, "output_throughput": { "median": 14.67, "min": 14.36, "max": 15.0, "n": 3 }, "median_ttft_ms": { "median": 204.28, "min": 203.37, "max": 204.78, "n": 3 }, "p99_ttft_ms": { "median": 587.75, "min": 548.55, "max": 606.37, "n": 3 }, "median_tpot_ms": { "median": 60.64, "min": 60.62, "max": 60.68, "n": 3 }, "p99_tpot_ms": { "median": 89.94, "min": 81.01, "max": 102.16, "n": 3 }, "p99_itl_ms": { "median": 103.64, "min": 103.15, "max": 104.32, "n": 3 }, "median_itl_ms": { "median": 60.89, "min": 60.85, "max": 60.91, "n": 3 }, "goodput_rps": { "median": 0.11, "min": 0.11, "max": 0.12, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.11, "max": 0.12, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1290.0, "max": 1335.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.412 }, "ttft_quantiles_ms": [ 90.8, 90.8, 91.6, 92.0, 184.5, 188.1, 192.4, 193.6, 194.5, 196.0, 197.0, 197.0, 197.3, 197.9, 198.0, 198.5, 199.2, 199.2, 199.5, 199.9, 200.7, 201.3, 202.2, 202.6, 204.2, 204.3, 204.4, 204.5, 205.0, 207.2, 208.2, 208.4, 221.3, 318.8, 318.8, 322.2, 332.8, 378.5, 392.3, 401.9, 459.4, 462.7, 463.2, 470.0, 474.3, 476.6, 476.8, 486.9, 561.7, 609.7, 627.5 ], "tpot_quantiles_ms": [ 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.5, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.6, 60.7, 60.7, 60.7, 60.7, 60.8, 60.9, 61.0, 61.0, 61.1, 61.1, 61.2, 68.1, 68.8, 76.0, 76.6, 78.2, 80.9, 81.4, 81.5, 82.0, 82.9, 91.2, 95.2, 103.4 ] }, { "x": 2.0, "output_throughput": { "median": 30.83, "min": 30.82, "max": 30.84, "n": 3 }, "median_ttft_ms": { "median": 126.46, "min": 125.89, "max": 134.04, "n": 3 }, "p99_ttft_ms": { "median": 155.75, "min": 155.67, "max": 156.29, "n": 3 }, "median_tpot_ms": { "median": 64.34, "min": 64.34, "max": 64.35, "n": 3 }, "p99_tpot_ms": { "median": 64.64, "min": 64.63, "max": 64.67, "n": 3 }, "p99_itl_ms": { "median": 66.2, "min": 66.18, "max": 66.41, "n": 3 }, "median_itl_ms": { "median": 64.26, "min": 64.24, "max": 64.27, "n": 3 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.24, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1515.0, "max": 1515.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.9, 95.9, 96.1, 96.7, 96.9, 97.2, 97.2, 97.2, 97.2, 97.4, 97.6, 97.6, 97.6, 97.6, 97.7, 97.8, 98.0, 98.0, 98.2, 98.2, 98.3, 98.3, 98.5, 99.4, 99.4, 114.5, 152.4, 152.6, 153.0, 153.5, 153.6, 153.6, 153.7, 153.7, 153.7, 153.7, 153.8, 153.9, 153.9, 154.0, 154.2, 154.3, 154.7, 154.7, 154.7, 154.7, 154.7, 155.6, 155.7, 155.9, 156.6 ], "tpot_quantiles_ms": [ 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.6, 64.7, 64.7, 64.7 ] }, { "x": 4.0, "output_throughput": { "median": 60.83, "min": 60.82, "max": 60.84, "n": 3 }, "median_ttft_ms": { "median": 159.13, "min": 158.81, "max": 160.21, "n": 3 }, "p99_ttft_ms": { "median": 161.58, "min": 161.34, "max": 174.08, "n": 3 }, "median_tpot_ms": { "median": 64.95, "min": 64.95, "max": 64.96, "n": 3 }, "p99_tpot_ms": { "median": 65.44, "min": 65.43, "max": 65.44, "n": 3 }, "p99_itl_ms": { "median": 66.7, "min": 66.5, "max": 68.09, "n": 3 }, "median_itl_ms": { "median": 65.06, "min": 65.02, "max": 65.08, "n": 3 }, "goodput_rps": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "request_throughput": { "median": 0.48, "min": 0.48, "max": 0.48, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1560.0, "min": 1560.0, "max": 1575.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 98.9, 98.9, 99.2, 99.6, 100.0, 100.2, 100.9, 100.9, 101.0, 101.6, 102.0, 102.8, 118.0, 157.2, 157.9, 158.2, 158.6, 158.6, 158.6, 158.6, 158.7, 158.9, 159.0, 159.1, 159.2, 159.2, 159.3, 159.5, 159.5, 159.6, 159.6, 159.6, 159.7, 159.9, 159.9, 160.0, 160.2, 160.2, 160.3, 160.3, 160.4, 160.5, 160.6, 160.7, 160.8, 161.0, 161.0, 161.1, 161.4, 161.7, 176.5 ], "tpot_quantiles_ms": [ 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.3, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4 ] }, { "x": 8.0, "output_throughput": { "median": 111.67, "min": 111.45, "max": 111.9, "n": 3 }, "median_ttft_ms": { "median": 185.54, "min": 183.13, "max": 189.75, "n": 3 }, "p99_ttft_ms": { "median": 1262.73, "min": 1219.41, "max": 1296.89, "n": 3 }, "median_tpot_ms": { "median": 66.6, "min": 66.58, "max": 66.83, "n": 3 }, "p99_tpot_ms": { "median": 74.8, "min": 74.51, "max": 75.37, "n": 3 }, "p99_itl_ms": { "median": 68.15, "min": 67.69, "max": 68.57, "n": 3 }, "median_itl_ms": { "median": 66.65, "min": 66.64, "max": 66.7, "n": 3 }, "goodput_rps": { "median": 0.49, "min": 0.49, "max": 0.63, "n": 3 }, "request_throughput": { "median": 0.87, "min": 0.87, "max": 0.87, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1530.0, "min": 1500.0, "max": 1545.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 100.0, 100.3, 102.0, 113.1, 158.0, 158.4, 158.7, 159.0, 159.1, 159.7, 160.1, 160.2, 161.0, 161.1, 162.1, 162.5, 162.8, 163.1, 163.3, 163.7, 164.0, 164.1, 173.3, 173.8, 174.5, 175.2, 204.3, 208.8, 228.1, 932.3, 933.6, 934.2, 1049.4, 1050.8, 1052.3, 1065.3, 1065.9, 1067.6, 1104.9, 1106.3, 1107.8, 1160.7, 1161.7, 1163.1, 1211.9, 1212.9, 1214.3, 1215.6, 1225.9, 1234.9, 1328.8 ], "tpot_quantiles_ms": [ 66.3, 66.3, 66.4, 66.4, 66.4, 66.4, 66.4, 66.4, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.5, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.6, 66.8, 67.0, 67.0, 67.1, 67.8, 67.8, 67.8, 67.9, 67.9, 67.9, 67.9, 67.9, 67.9, 68.1, 68.1, 68.1, 69.0, 69.0, 69.0, 74.5, 74.7, 75.7 ] }, { "x": 16.0, "output_throughput": { "median": 200.16, "min": 199.41, "max": 200.19, "n": 3 }, "median_ttft_ms": { "median": 234.36, "min": 233.33, "max": 237.58, "n": 3 }, "p99_ttft_ms": { "median": 2411.08, "min": 2402.58, "max": 2438.6, "n": 3 }, "median_tpot_ms": { "median": 70.59, "min": 70.5, "max": 70.97, "n": 3 }, "p99_tpot_ms": { "median": 87.48, "min": 87.46, "max": 87.76, "n": 3 }, "p99_itl_ms": { "median": 72.83, "min": 72.07, "max": 73.64, "n": 3 }, "median_itl_ms": { "median": 70.23, "min": 70.15, "max": 70.3, "n": 3 }, "goodput_rps": { "median": 0.95, "min": 0.83, "max": 1.08, "n": 3 }, "request_throughput": { "median": 1.56, "min": 1.56, "max": 1.56, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1440.0, "max": 1455.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 103.4, 106.8, 164.1, 164.8, 165.6, 167.0, 168.6, 173.2, 174.0, 174.9, 175.5, 176.3, 177.4, 203.1, 221.5, 222.2, 223.4, 224.0, 224.7, 225.7, 227.9, 232.4, 233.2, 233.5, 234.5, 236.1, 237.4, 922.9, 927.0, 937.3, 954.5, 1053.1, 1056.3, 1058.3, 1230.1, 1231.9, 2102.9, 2105.1, 2106.2, 2107.2, 2112.2, 2114.5, 2203.0, 2205.4, 2207.5, 2209.2, 2210.8, 2395.0, 2406.1, 2408.7, 2488.9 ], "tpot_quantiles_ms": [ 69.9, 69.9, 70.0, 70.0, 70.0, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.1, 70.2, 70.4, 70.4, 70.4, 70.4, 70.5, 70.5, 70.5, 70.5, 70.5, 70.6, 70.6, 70.6, 70.9, 70.9, 71.1, 71.1, 71.1, 71.5, 71.5, 71.5, 71.5, 72.4, 72.5, 72.5, 72.5, 72.6, 72.6, 80.4, 80.4, 80.6, 80.6, 80.6, 81.5, 81.5, 81.8, 81.9, 87.3, 88.5 ] }, { "x": 32.0, "output_throughput": { "median": 270.81, "min": 269.91, "max": 271.99, "n": 3 }, "median_ttft_ms": { "median": 3122.98, "min": 3118.11, "max": 3229.92, "n": 3 }, "p99_ttft_ms": { "median": 4810.53, "min": 4742.83, "max": 4826.08, "n": 3 }, "median_tpot_ms": { "median": 94.12, "min": 94.06, "max": 94.16, "n": 3 }, "p99_tpot_ms": { "median": 117.95, "min": 117.4, "max": 117.97, "n": 3 }, "p99_itl_ms": { "median": 87.21, "min": 86.76, "max": 87.9, "n": 3 }, "median_itl_ms": { "median": 82.17, "min": 82.07, "max": 82.19, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.12, "min": 2.11, "max": 2.12, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1320.0, "max": 1350.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 116.6, 211.1, 808.8, 934.4, 940.5, 946.7, 957.1, 1067.9, 1072.3, 1076.6, 1100.5, 1965.8, 1967.0, 2089.8, 2114.7, 2118.9, 2130.0, 2157.9, 2212.9, 2216.2, 2245.0, 2252.2, 2255.9, 3116.6, 3121.5, 3227.6, 3279.9, 3281.8, 3301.0, 3343.6, 3352.5, 3360.1, 3392.0, 3394.7, 3423.0, 3429.7, 4266.7, 4270.6, 4378.2, 4416.3, 4449.9, 4455.6, 4484.7, 4505.9, 4515.9, 4544.0, 4599.8, 4602.6, 4620.2, 4742.6, 4848.1 ], "tpot_quantiles_ms": [ 82.0, 82.0, 82.0, 82.2, 82.3, 83.5, 83.5, 83.5, 83.8, 83.9, 84.4, 84.5, 84.6, 84.6, 85.1, 85.1, 90.1, 91.0, 92.6, 92.8, 92.8, 92.9, 93.2, 93.5, 93.8, 94.1, 94.2, 94.2, 99.1, 100.0, 101.5, 101.9, 102.0, 102.0, 102.4, 102.5, 103.0, 103.2, 103.4, 103.5, 108.0, 109.1, 110.6, 111.1, 111.3, 111.6, 112.3, 112.3, 112.9, 117.3, 118.8 ] }, { "x": 64.0, "output_throughput": { "median": 265.3, "min": 260.15, "max": 268.21, "n": 3 }, "median_ttft_ms": { "median": 18333.46, "min": 18262.52, "max": 18378.46, "n": 3 }, "p99_ttft_ms": { "median": 29410.5, "min": 28625.25, "max": 31260.17, "n": 3 }, "median_tpot_ms": { "median": 103.12, "min": 102.7, "max": 105.08, "n": 3 }, "p99_tpot_ms": { "median": 150.42, "min": 146.54, "max": 154.98, "n": 3 }, "p99_itl_ms": { "median": 90.03, "min": 89.43, "max": 92.71, "n": 3 }, "median_itl_ms": { "median": 84.1, "min": 84.03, "max": 84.19, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.07, "min": 2.03, "max": 2.1, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1260.0, "max": 1290.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.322 }, "ttft_quantiles_ms": [ 185.3, 1690.0, 1840.8, 2993.7, 3977.7, 4150.0, 4644.9, 4678.4, 4752.4, 5292.8, 5308.7, 5371.3, 5392.1, 5446.1, 6161.0, 7350.5, 9078.7, 10219.6, 11074.0, 17181.5, 17220.5, 17234.0, 17248.3, 17324.3, 17412.8, 18264.3, 18366.9, 18381.0, 18416.9, 18578.5, 18638.0, 19062.9, 19414.5, 19518.1, 19556.9, 19590.3, 19756.5, 19987.9, 20464.3, 20709.8, 20768.7, 21054.8, 22333.6, 22992.6, 23097.1, 24720.5, 25316.1, 25666.2, 27475.5, 30029.4, 32725.2 ], "tpot_quantiles_ms": [ 83.5, 84.1, 84.2, 84.2, 84.3, 84.5, 84.7, 85.2, 86.1, 89.6, 89.8, 89.9, 90.0, 90.3, 91.2, 92.2, 93.9, 94.4, 95.3, 98.9, 99.0, 99.1, 99.5, 102.2, 102.7, 103.4, 106.2, 106.9, 108.0, 108.1, 108.2, 108.4, 108.9, 111.8, 112.2, 113.3, 115.1, 117.1, 117.2, 117.3, 118.0, 119.4, 121.7, 123.2, 126.1, 128.8, 132.1, 138.7, 145.7, 147.2, 155.0 ] } ] } }, "rate": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 70.62, "min": 67.21, "max": 92.27, "n": 3 }, "median_ttft_ms": { "median": 276.06, "min": 275.56, "max": 290.69, "n": 3 }, "p99_ttft_ms": { "median": 428.06, "min": 408.66, "max": 478.95, "n": 3 }, "median_tpot_ms": { "median": 74.73, "min": 72.23, "max": 76.35, "n": 3 }, "p99_tpot_ms": { "median": 80.77, "min": 80.32, "max": 83.51, "n": 3 }, "goodput_rps": { "median": 0.55, "min": 0.53, "max": 0.72, "n": 3 }, "request_throughput": { "median": 0.55, "min": 0.53, "max": 0.72, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1425.0, "max": 1500.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 71.1, 71.6, 242.6, 245.5, 248.8, 249.9, 250.9, 253.5, 254.3, 255.2, 260.1, 260.5, 261.9, 263.4, 266.8, 267.7, 269.2, 269.5, 272.3, 273.4, 274.4, 275.0, 277.3, 278.8, 280.1, 281.6, 282.7, 286.2, 286.9, 289.2, 291.1, 291.7, 293.0, 294.7, 298.5, 299.5, 301.3, 302.0, 305.1, 306.2, 308.4, 310.2, 321.5, 332.9, 352.7, 372.3, 389.9, 403.0, 425.0, 460.9, 486.7 ], "tpot_quantiles_ms": [ 60.4, 63.4, 65.4, 66.0, 66.4, 66.7, 67.3, 68.1, 68.1, 68.3, 68.7, 68.8, 69.4, 69.6, 69.7, 71.0, 71.2, 71.5, 72.0, 72.4, 72.7, 73.0, 73.0, 73.5, 73.9, 74.1, 74.7, 74.7, 75.0, 75.5, 76.0, 76.0, 76.1, 76.2, 76.4, 76.5, 76.9, 77.3, 77.6, 77.6, 77.9, 77.9, 78.6, 78.9, 79.2, 79.4, 80.1, 80.3, 80.5, 81.8, 83.9 ] }, { "x": 1.0, "output_throughput": { "median": 123.3, "min": 118.21, "max": 153.55, "n": 3 }, "median_ttft_ms": { "median": 174.73, "min": 167.7, "max": 184.73, "n": 3 }, "p99_ttft_ms": { "median": 223.24, "min": 207.88, "max": 228.02, "n": 3 }, "median_tpot_ms": { "median": 74.02, "min": 72.51, "max": 74.53, "n": 3 }, "p99_tpot_ms": { "median": 76.75, "min": 76.17, "max": 77.73, "n": 3 }, "goodput_rps": { "median": 0.96, "min": 0.92, "max": 1.2, "n": 3 }, "request_throughput": { "median": 0.96, "min": 0.92, "max": 1.2, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1485.0, "max": 1500.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 70.7, 71.7, 115.0, 126.1, 135.9, 137.6, 140.2, 140.7, 141.8, 144.4, 147.8, 148.0, 149.4, 152.3, 154.8, 156.6, 158.8, 160.5, 163.1, 164.6, 166.6, 167.0, 168.4, 169.6, 170.2, 173.6, 174.8, 176.0, 177.4, 180.5, 184.9, 185.0, 185.8, 186.9, 187.9, 188.5, 188.8, 189.2, 191.8, 193.4, 194.4, 194.8, 195.9, 196.4, 198.5, 200.6, 203.3, 204.1, 209.6, 222.6, 230.5 ], "tpot_quantiles_ms": [ 62.4, 65.3, 65.9, 67.3, 67.7, 68.1, 68.5, 68.8, 69.2, 69.5, 69.9, 70.1, 70.1, 70.4, 70.8, 71.1, 71.2, 71.7, 72.3, 72.5, 72.8, 73.0, 73.1, 73.5, 73.5, 73.8, 73.8, 74.0, 74.0, 74.0, 74.2, 74.2, 74.5, 74.7, 74.9, 75.0, 75.2, 75.3, 75.6, 75.7, 76.0, 76.1, 76.2, 76.3, 76.5, 76.6, 76.6, 76.7, 76.9, 77.4, 77.8 ] }, { "x": 2.0, "output_throughput": { "median": 226.44, "min": 215.48, "max": 229.95, "n": 3 }, "median_ttft_ms": { "median": 223.21, "min": 213.73, "max": 231.08, "n": 3 }, "p99_ttft_ms": { "median": 623.14, "min": 472.02, "max": 1445.31, "n": 3 }, "median_tpot_ms": { "median": 100.61, "min": 93.18, "max": 105.84, "n": 3 }, "p99_tpot_ms": { "median": 121.6, "min": 117.26, "max": 122.17, "n": 3 }, "goodput_rps": { "median": 0.88, "min": 0.51, "max": 1.15, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.68, "max": 1.8, "n": 3 }, "completed": [ 60, 60, 60 ], "gpu": { "median_sm_mhz": { "median": 1425.0, "min": 1417.5, "max": 1455.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 69.8, 106.5, 131.3, 139.4, 145.2, 148.6, 152.3, 157.4, 158.9, 162.1, 164.5, 167.7, 169.9, 172.1, 174.3, 179.5, 184.5, 186.7, 194.4, 198.0, 201.0, 202.6, 205.2, 210.1, 216.5, 224.2, 235.7, 254.7, 274.0, 283.3, 294.9, 299.1, 305.0, 312.8, 316.9, 326.1, 330.8, 332.4, 339.0, 348.7, 365.1, 374.5, 392.1, 399.4, 407.3, 431.0, 439.2, 462.5, 618.8, 1176.9, 1638.9 ], "tpot_quantiles_ms": [ 71.0, 76.6, 78.9, 81.0, 82.5, 83.6, 84.1, 84.4, 84.9, 85.6, 85.9, 86.0, 86.3, 88.6, 89.3, 90.0, 91.6, 91.6, 93.2, 93.9, 94.7, 95.5, 95.9, 97.2, 98.2, 99.9, 100.4, 101.0, 101.9, 102.8, 104.1, 105.5, 106.9, 108.0, 108.7, 109.6, 111.0, 112.3, 113.8, 114.9, 116.2, 116.7, 117.0, 118.0, 119.3, 119.7, 120.3, 120.7, 121.0, 121.7, 122.2 ] }, { "x": 4.0, "output_throughput": { "median": 256.94, "min": 254.46, "max": 259.32, "n": 3 }, "median_ttft_ms": { "median": 9640.27, "min": 8454.34, "max": 9930.61, "n": 3 }, "p99_ttft_ms": { "median": 24331.32, "min": 21608.36, "max": 25467.4, "n": 3 }, "median_tpot_ms": { "median": 115.08, "min": 114.19, "max": 117.41, "n": 3 }, "p99_tpot_ms": { "median": 127.58, "min": 126.31, "max": 128.29, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.01, "min": 1.99, "max": 2.03, "n": 3 }, "completed": [ 120, 120, 120 ], "gpu": { "median_sm_mhz": { "median": 1320.0, "min": 1305.0, "max": 1335.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 71.5, 268.9, 321.4, 385.2, 421.8, 520.9, 599.6, 653.7, 717.8, 769.2, 826.7, 866.8, 929.9, 1045.6, 1305.7, 1691.3, 7013.9, 7598.4, 7753.1, 7923.7, 8184.5, 8466.3, 8598.0, 8934.9, 9089.2, 9304.6, 9511.0, 9899.5, 10114.4, 10392.1, 10578.6, 12962.4, 14304.0, 14593.2, 15051.1, 15451.9, 15812.9, 16197.3, 16307.6, 16516.4, 16642.8, 16834.9, 17078.6, 17374.9, 17691.6, 18641.7, 20178.3, 21213.2, 23913.0, 24298.9, 25532.8 ], "tpot_quantiles_ms": [ 66.4, 68.5, 72.7, 76.5, 92.4, 95.9, 99.3, 100.3, 100.3, 101.0, 102.1, 107.0, 109.3, 109.6, 110.5, 111.2, 111.8, 113.5, 113.6, 113.7, 114.2, 114.8, 115.1, 115.2, 115.7, 115.7, 116.7, 117.4, 118.9, 119.0, 119.7, 120.0, 120.0, 121.1, 121.4, 122.0, 122.8, 123.8, 123.9, 124.7, 125.1, 125.3, 125.9, 126.0, 126.1, 126.6, 126.9, 127.2, 127.5, 127.8, 128.6 ] }, { "x": 8.0, "output_throughput": { "median": 244.12, "min": 243.24, "max": 244.64, "n": 3 }, "median_ttft_ms": { "median": 37438.69, "min": 35626.32, "max": 37473.65, "n": 3 }, "p99_ttft_ms": { "median": 87898.19, "min": 83677.56, "max": 87917.71, "n": 3 }, "median_tpot_ms": { "median": 108.25, "min": 107.23, "max": 108.69, "n": 3 }, "p99_tpot_ms": { "median": 166.72, "min": 160.22, "max": 174.69, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.91, "min": 1.9, "max": 1.91, "n": 3 }, "completed": [ 240, 240, 240 ], "gpu": { "median_sm_mhz": { "median": 1102.5, "min": 1065.0, "max": 1177.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.378 }, "ttft_quantiles_ms": [ 70.6, 724.8, 1061.4, 1562.6, 1754.9, 1950.2, 2183.6, 2617.4, 12378.4, 13149.9, 13546.0, 13810.4, 13949.4, 14244.1, 14483.0, 14644.7, 24656.2, 25024.3, 25324.7, 25606.9, 25861.8, 26108.3, 26392.7, 26804.8, 35719.4, 36099.5, 36627.0, 37474.8, 37770.1, 38240.8, 38456.8, 46858.9, 48738.5, 49765.7, 50365.3, 50707.7, 51338.6, 52470.8, 53717.4, 63658.7, 64907.9, 66166.9, 67110.4, 67835.2, 69757.7, 71103.8, 72144.7, 82711.6, 85080.9, 87456.6, 88345.8 ], "tpot_quantiles_ms": [ 85.3, 85.4, 85.5, 86.9, 87.5, 88.7, 88.8, 89.3, 89.9, 90.5, 92.5, 93.9, 94.5, 94.8, 95.7, 97.1, 97.3, 98.2, 98.3, 99.3, 99.8, 103.2, 104.1, 105.2, 107.0, 107.6, 108.2, 108.8, 108.9, 109.3, 112.9, 113.6, 115.0, 115.5, 116.5, 117.0, 118.1, 118.3, 120.3, 121.4, 122.1, 124.4, 125.6, 126.1, 134.7, 137.2, 140.2, 145.5, 158.4, 166.7, 184.4 ] }, { "x": 16.0, "output_throughput": { "median": 206.85, "min": 206.51, "max": 207.29, "n": 3 }, "median_ttft_ms": { "median": 103234.8, "min": 101465.74, "max": 104099.81, "n": 3 }, "p99_ttft_ms": { "median": 254781.08, "min": 251615.66, "max": 255294.31, "n": 3 }, "median_tpot_ms": { "median": 117.09, "min": 116.27, "max": 120.07, "n": 3 }, "p99_tpot_ms": { "median": 200.69, "min": 199.29, "max": 205.01, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.62, "min": 1.61, "max": 1.62, "n": 3 }, "completed": [ 480, 480, 480 ], "gpu": { "median_sm_mhz": { "median": 465.0, "min": 450.0, "max": 510.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.702 }, "ttft_quantiles_ms": [ 71.4, 1563.7, 2794.8, 3546.9, 14933.8, 15957.9, 16839.3, 17711.1, 29341.4, 30170.0, 30966.1, 31655.0, 43252.9, 44213.9, 45016.4, 45932.9, 58028.6, 59680.8, 61295.0, 63275.8, 76635.7, 79255.3, 81528.9, 84491.2, 99357.9, 101972.1, 104336.1, 107390.1, 123033.9, 126163.9, 128326.9, 143781.7, 147873.9, 150692.2, 153150.6, 169257.9, 174097.1, 176526.6, 178471.3, 194815.4, 199072.5, 201634.8, 204321.4, 221591.6, 224520.4, 227419.6, 229718.3, 247218.4, 249909.6, 252621.5, 255973.6 ], "tpot_quantiles_ms": [ 83.7, 84.0, 84.6, 88.4, 88.5, 89.3, 92.8, 96.7, 97.7, 98.0, 98.4, 99.3, 102.0, 105.1, 106.1, 106.7, 107.2, 107.5, 108.3, 110.5, 111.6, 112.7, 115.4, 116.5, 116.8, 117.2, 119.6, 121.9, 124.8, 127.7, 129.1, 131.6, 135.5, 136.6, 137.6, 142.8, 145.3, 149.5, 153.2, 156.2, 158.1, 160.3, 161.7, 166.4, 169.8, 178.5, 179.7, 181.8, 186.5, 187.6, 219.5 ] } ] }, "fp8": { "kv_pool_tokens": 142369, "points": [ { "x": 0.5, "output_throughput": { "median": 74.79, "min": 70.39, "max": 99.72, "n": 3 }, "median_ttft_ms": { "median": 210.5, "min": 206.41, "max": 210.51, "n": 3 }, "p99_ttft_ms": { "median": 302.78, "min": 291.34, "max": 350.95, "n": 3 }, "median_tpot_ms": { "median": 43.9, "min": 43.85, "max": 45.53, "n": 3 }, "p99_tpot_ms": { "median": 48.95, "min": 48.89, "max": 51.12, "n": 3 }, "goodput_rps": { "median": 0.58, "min": 0.55, "max": 0.78, "n": 3 }, "request_throughput": { "median": 0.58, "min": 0.55, "max": 0.78, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1245.0, "min": 1245.0, "max": 1275.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 61.7, 62.3, 131.2, 138.0, 181.9, 184.4, 185.0, 185.8, 186.4, 188.7, 189.1, 189.3, 190.0, 191.8, 193.2, 194.6, 196.8, 197.2, 197.3, 198.8, 202.9, 204.4, 205.5, 207.3, 207.8, 208.8, 209.0, 210.2, 211.1, 212.9, 212.9, 213.0, 213.5, 214.1, 215.4, 217.7, 218.3, 220.1, 221.2, 221.7, 222.5, 222.7, 224.6, 225.9, 226.6, 235.0, 271.1, 272.6, 299.8, 309.9, 351.2 ], "tpot_quantiles_ms": [ 39.7, 39.7, 39.8, 40.4, 40.8, 40.9, 41.2, 41.3, 41.9, 41.9, 42.3, 42.3, 42.4, 42.6, 42.8, 43.4, 43.5, 43.5, 43.6, 43.6, 43.8, 43.8, 43.9, 44.0, 44.5, 44.6, 44.6, 44.7, 44.7, 44.8, 44.9, 44.9, 45.0, 45.5, 45.6, 45.7, 45.7, 45.8, 46.0, 46.1, 46.5, 46.7, 47.0, 47.1, 48.0, 48.2, 48.9, 49.1, 49.9, 50.2, 51.2 ] }, { "x": 1.0, "output_throughput": { "median": 136.78, "min": 129.4, "max": 176.18, "n": 3 }, "median_ttft_ms": { "median": 119.21, "min": 112.15, "max": 119.74, "n": 3 }, "p99_ttft_ms": { "median": 166.64, "min": 164.78, "max": 172.27, "n": 3 }, "median_tpot_ms": { "median": 44.14, "min": 42.88, "max": 44.21, "n": 3 }, "p99_tpot_ms": { "median": 45.26, "min": 45.12, "max": 45.81, "n": 3 }, "goodput_rps": { "median": 1.07, "min": 1.01, "max": 1.38, "n": 3 }, "request_throughput": { "median": 1.07, "min": 1.01, "max": 1.38, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1305.0, "max": 1365.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 59.8, 60.9, 62.1, 67.7, 91.2, 94.3, 98.0, 98.7, 100.1, 100.8, 101.9, 102.7, 104.1, 105.5, 107.3, 108.3, 109.5, 109.6, 109.7, 110.1, 111.0, 111.3, 112.8, 113.1, 114.7, 116.2, 118.3, 120.2, 122.3, 123.5, 123.6, 123.8, 124.2, 125.1, 125.9, 126.6, 127.2, 127.7, 128.7, 130.5, 131.3, 132.0, 133.3, 135.9, 139.4, 143.0, 156.6, 162.1, 166.0, 169.7, 173.5 ], "tpot_quantiles_ms": [ 39.6, 40.2, 40.2, 40.9, 41.2, 41.3, 41.3, 41.5, 41.5, 41.8, 41.8, 41.8, 41.9, 42.1, 42.2, 42.4, 42.5, 42.8, 43.0, 43.1, 43.1, 43.1, 43.2, 43.3, 43.5, 43.6, 43.7, 43.7, 43.9, 44.1, 44.1, 44.2, 44.2, 44.4, 44.4, 44.4, 44.5, 44.5, 44.6, 44.6, 44.7, 44.7, 44.8, 44.8, 45.1, 45.1, 45.2, 45.3, 45.3, 45.5, 45.8 ] }, { "x": 2.0, "output_throughput": { "median": 259.98, "min": 245.31, "max": 260.81, "n": 3 }, "median_ttft_ms": { "median": 166.99, "min": 163.39, "max": 172.68, "n": 3 }, "p99_ttft_ms": { "median": 383.74, "min": 302.66, "max": 401.73, "n": 3 }, "median_tpot_ms": { "median": 49.94, "min": 49.09, "max": 54.85, "n": 3 }, "p99_tpot_ms": { "median": 61.16, "min": 59.9, "max": 70.79, "n": 3 }, "goodput_rps": { "median": 2.03, "min": 1.92, "max": 2.04, "n": 3 }, "request_throughput": { "median": 2.03, "min": 1.92, "max": 2.04, "n": 3 }, "completed": [ 60, 60, 60 ], "gpu": { "median_sm_mhz": { "median": 1260.0, "min": 1260.0, "max": 1260.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 59.4, 65.7, 96.6, 99.8, 101.4, 102.7, 106.1, 109.8, 110.2, 111.5, 114.7, 117.9, 119.4, 121.6, 123.9, 124.4, 125.9, 127.0, 131.5, 136.5, 137.5, 140.6, 142.0, 150.9, 162.6, 167.1, 171.4, 183.7, 189.4, 191.9, 195.0, 196.2, 202.7, 205.4, 208.0, 209.7, 210.2, 213.3, 216.5, 218.5, 221.7, 222.5, 224.6, 229.2, 239.2, 255.1, 275.6, 296.9, 307.1, 334.5, 422.5 ], "tpot_quantiles_ms": [ 40.6, 43.0, 44.6, 45.3, 45.6, 45.8, 45.9, 46.5, 46.7, 46.9, 47.1, 47.2, 47.4, 47.9, 48.1, 48.2, 48.5, 48.8, 48.9, 49.2, 49.7, 49.9, 50.0, 50.3, 50.7, 50.9, 51.3, 51.5, 52.1, 52.4, 52.6, 53.4, 53.7, 54.3, 54.8, 55.2, 56.0, 56.4, 57.5, 58.0, 58.6, 58.7, 58.8, 59.1, 59.8, 59.9, 60.8, 63.0, 66.0, 68.0, 70.9 ] }, { "x": 4.0, "output_throughput": { "median": 465.88, "min": 421.38, "max": 469.18, "n": 3 }, "median_ttft_ms": { "median": 188.86, "min": 188.53, "max": 193.86, "n": 3 }, "p99_ttft_ms": { "median": 480.41, "min": 417.88, "max": 828.09, "n": 3 }, "median_tpot_ms": { "median": 70.79, "min": 67.99, "max": 81.45, "n": 3 }, "p99_tpot_ms": { "median": 111.49, "min": 95.66, "max": 112.96, "n": 3 }, "goodput_rps": { "median": 3.18, "min": 2.93, "max": 3.29, "n": 3 }, "request_throughput": { "median": 3.64, "min": 3.29, "max": 3.67, "n": 3 }, "completed": [ 120, 120, 120 ], "gpu": { "median_sm_mhz": { "median": 1252.5, "min": 1230.0, "max": 1365.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 60.5, 66.6, 79.3, 89.4, 98.8, 103.2, 108.1, 111.0, 115.1, 118.1, 121.1, 123.6, 125.6, 128.4, 131.1, 134.6, 138.0, 142.2, 144.4, 146.0, 149.8, 153.9, 162.5, 171.1, 178.7, 187.4, 198.9, 203.0, 208.9, 213.6, 217.5, 221.2, 226.0, 229.0, 237.0, 242.1, 253.4, 271.1, 284.9, 298.5, 302.2, 316.5, 322.7, 328.7, 339.9, 371.1, 383.8, 405.7, 435.0, 496.8, 857.9 ], "tpot_quantiles_ms": [ 42.8, 48.1, 50.6, 52.2, 52.8, 53.2, 53.4, 53.8, 55.0, 56.1, 57.3, 58.0, 58.5, 59.3, 59.6, 59.9, 60.3, 60.6, 60.6, 61.0, 61.8, 64.3, 65.7, 68.6, 69.9, 71.4, 72.9, 75.4, 78.1, 80.5, 82.1, 82.8, 84.7, 86.0, 88.3, 90.1, 90.7, 92.1, 92.7, 93.5, 95.0, 95.3, 95.7, 96.8, 99.2, 100.2, 101.2, 104.3, 106.1, 110.0, 114.2 ] }, { "x": 8.0, "output_throughput": { "median": 728.61, "min": 716.03, "max": 736.28, "n": 3 }, "median_ttft_ms": { "median": 270.08, "min": 195.41, "max": 297.19, "n": 3 }, "p99_ttft_ms": { "median": 2923.87, "min": 1858.39, "max": 3022.89, "n": 3 }, "median_tpot_ms": { "median": 182.1, "min": 169.75, "max": 188.6, "n": 3 }, "p99_tpot_ms": { "median": 220.47, "min": 212.41, "max": 223.02, "n": 3 }, "goodput_rps": { "median": 0.84, "min": 0.62, "max": 1.1, "n": 3 }, "request_throughput": { "median": 5.69, "min": 5.59, "max": 5.75, "n": 3 }, "completed": [ 240, 240, 240 ], "gpu": { "median_sm_mhz": { "median": 1110.0, "min": 1087.5, "max": 1125.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 54.2, 82.2, 101.4, 105.5, 109.3, 112.0, 115.7, 117.9, 122.4, 124.6, 126.5, 129.5, 132.0, 133.9, 137.5, 140.5, 143.2, 147.2, 150.4, 153.4, 158.0, 161.4, 165.1, 170.1, 177.9, 262.1, 337.1, 430.3, 505.0, 557.2, 590.5, 655.5, 712.7, 801.8, 871.3, 965.6, 1132.6, 1224.5, 1290.2, 1457.1, 1554.8, 1616.8, 1693.0, 1859.2, 2023.7, 2158.0, 2260.4, 2427.0, 2583.4, 2758.9, 3140.8 ], "tpot_quantiles_ms": [ 73.1, 74.7, 80.7, 82.9, 84.2, 85.5, 86.9, 89.2, 100.1, 104.7, 108.2, 113.6, 116.7, 122.5, 129.4, 132.3, 136.3, 141.8, 147.5, 151.9, 155.2, 161.6, 166.0, 170.5, 175.6, 178.4, 185.2, 187.2, 193.9, 198.1, 202.3, 204.1, 204.4, 205.4, 207.0, 208.5, 209.3, 211.7, 212.0, 214.1, 217.4, 218.5, 218.9, 219.4, 219.6, 219.8, 220.5, 221.5, 222.0, 222.4, 223.5 ] }, { "x": 16.0, "output_throughput": { "median": 718.5, "min": 717.43, "max": 739.86, "n": 3 }, "median_ttft_ms": { "median": 12345.92, "min": 9856.86, "max": 13563.28, "n": 3 }, "p99_ttft_ms": { "median": 48094.39, "min": 43765.57, "max": 49885.65, "n": 3 }, "median_tpot_ms": { "median": 259.32, "min": 248.3, "max": 269.01, "n": 3 }, "p99_tpot_ms": { "median": 358.01, "min": 344.75, "max": 360.47, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 5.61, "min": 5.6, "max": 5.78, "n": 3 }, "completed": [ 480, 480, 480 ], "gpu": { "median_sm_mhz": { "median": 1005.0, "min": 982.5, "max": 1005.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.306 }, "ttft_quantiles_ms": [ 59.5, 109.8, 119.3, 129.3, 137.5, 142.8, 149.3, 154.8, 163.5, 171.5, 178.8, 196.4, 219.7, 403.4, 796.1, 1191.0, 1438.8, 1667.3, 1903.2, 2059.4, 2322.7, 2539.8, 2916.9, 4110.5, 10198.4, 11271.2, 12195.4, 12972.1, 13414.3, 13949.4, 14752.2, 15332.6, 15938.6, 16760.9, 17504.6, 18225.9, 19182.5, 20069.3, 20789.2, 22147.2, 23092.6, 24134.1, 25340.3, 26646.6, 28007.5, 29015.6, 29964.3, 41850.6, 44897.2, 47809.3, 50327.4 ], "tpot_quantiles_ms": [ 54.4, 71.6, 85.8, 102.5, 134.7, 137.3, 139.5, 140.1, 140.2, 145.3, 145.6, 149.8, 157.4, 165.1, 171.8, 180.6, 191.2, 201.8, 214.9, 225.5, 232.7, 239.9, 244.1, 247.6, 253.0, 258.8, 263.4, 264.8, 269.5, 272.9, 275.4, 279.9, 282.4, 286.8, 287.9, 294.0, 295.7, 297.6, 303.2, 306.2, 310.6, 313.2, 318.0, 319.4, 321.4, 327.0, 334.6, 337.3, 343.0, 358.0, 366.0 ] } ] }, "dp2": { "kv_pool_tokens": 2925, "points": [ { "x": 0.5, "output_throughput": { "median": 50.19, "min": 46.27, "max": 54.72, "n": 3 }, "median_ttft_ms": { "median": 309.82, "min": 293.99, "max": 342.1, "n": 3 }, "p99_ttft_ms": { "median": 18017.76, "min": 12392.73, "max": 20757.17, "n": 3 }, "median_tpot_ms": { "median": 91.17, "min": 87.62, "max": 93.33, "n": 3 }, "p99_tpot_ms": { "median": 174.24, "min": 163.21, "max": 206.68, "n": 3 }, "p99_itl_ms": { "median": 287.21, "min": 286.37, "max": 358.83, "n": 3 }, "median_itl_ms": { "median": 65.07, "min": 64.95, "max": 65.09, "n": 3 }, "goodput_rps": { "median": 0.2, "min": 0.18, "max": 0.21, "n": 3 }, "request_throughput": { "median": 0.39, "min": 0.36, "max": 0.43, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1417.5, "max": 1500.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.224 }, "ttft_quantiles_ms": [ 163.5, 164.6, 170.8, 241.2, 247.9, 253.3, 256.5, 256.9, 260.0, 265.0, 269.9, 270.4, 271.3, 273.2, 274.3, 276.8, 277.6, 281.6, 283.5, 289.0, 291.4, 291.4, 292.5, 295.4, 297.0, 308.6, 322.0, 362.2, 421.7, 452.5, 499.9, 505.4, 520.3, 2641.0, 3329.2, 6401.3, 6854.2, 6957.4, 7462.6, 7852.1, 8408.3, 8803.9, 9466.1, 10957.0, 11327.8, 12138.2, 12425.3, 13501.4, 16199.5, 19744.7, 21212.0 ], "tpot_quantiles_ms": [ 60.1, 62.0, 63.2, 63.6, 64.4, 65.1, 65.3, 65.4, 66.7, 66.7, 66.9, 66.9, 66.9, 67.3, 67.7, 68.4, 68.4, 68.4, 68.5, 68.8, 69.4, 69.6, 69.8, 70.1, 71.3, 72.7, 110.5, 111.1, 112.4, 113.1, 113.1, 113.4, 114.1, 114.8, 115.4, 115.6, 116.9, 117.4, 117.8, 122.1, 125.2, 126.4, 138.4, 150.7, 153.2, 156.6, 158.9, 160.9, 181.1, 197.9, 209.3 ] }, { "x": 1.0, "output_throughput": { "median": 53.52, "min": 51.48, "max": 58.74, "n": 3 }, "median_ttft_ms": { "median": 309.42, "min": 299.42, "max": 353.35, "n": 3 }, "p99_ttft_ms": { "median": 32734.38, "min": 31217.42, "max": 33495.62, "n": 3 }, "median_tpot_ms": { "median": 94.2, "min": 94.11, "max": 94.9, "n": 3 }, "p99_tpot_ms": { "median": 163.04, "min": 157.1, "max": 199.2, "n": 3 }, "p99_itl_ms": { "median": 376.55, "min": 259.94, "max": 416.28, "n": 3 }, "median_itl_ms": { "median": 65.98, "min": 65.91, "max": 67.25, "n": 3 }, "goodput_rps": { "median": 0.21, "min": 0.2, "max": 0.23, "n": 3 }, "request_throughput": { "median": 0.42, "min": 0.4, "max": 0.46, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1515.0, "min": 1515.0, "max": 1552.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.236 }, "ttft_quantiles_ms": [ 73.1, 73.2, 175.1, 229.3, 245.8, 247.0, 251.2, 251.7, 260.1, 264.7, 267.8, 267.9, 270.9, 273.7, 275.4, 277.2, 283.2, 285.1, 286.3, 288.1, 292.9, 293.3, 297.5, 298.7, 301.2, 317.7, 351.7, 370.9, 426.5, 515.8, 527.9, 559.0, 7463.7, 9659.3, 11064.8, 11131.2, 11237.5, 11952.5, 17184.3, 18317.1, 18728.9, 18762.6, 20385.5, 20560.7, 21388.7, 29046.8, 31007.7, 31759.1, 32130.3, 33005.8, 33501.9 ], "tpot_quantiles_ms": [ 62.9, 63.2, 65.6, 67.0, 67.9, 68.2, 69.3, 69.7, 70.3, 71.2, 71.4, 71.4, 72.4, 72.6, 72.9, 73.3, 73.6, 74.0, 74.2, 74.4, 75.1, 75.2, 75.3, 75.8, 76.7, 77.9, 110.5, 112.1, 112.3, 112.5, 112.6, 112.9, 113.9, 114.5, 114.8, 115.4, 116.3, 116.8, 117.5, 121.9, 126.6, 130.6, 136.4, 149.1, 153.4, 159.5, 162.1, 162.8, 165.8, 187.2, 200.9 ] }, { "x": 2.0, "output_throughput": { "median": 44.61, "min": 41.08, "max": 46.25, "n": 3 }, "median_ttft_ms": { "median": 322.88, "min": 303.98, "max": 339.52, "n": 3 }, "p99_ttft_ms": { "median": 126800.34, "min": 119273.24, "max": 131845.35, "n": 3 }, "median_tpot_ms": { "median": 97.15, "min": 94.81, "max": 97.25, "n": 3 }, "p99_tpot_ms": { "median": 215.49, "min": 206.71, "max": 228.26, "n": 3 }, "p99_itl_ms": { "median": 355.89, "min": 248.76, "max": 388.67, "n": 3 }, "median_itl_ms": { "median": 69.6, "min": 68.85, "max": 75.26, "n": 3 }, "goodput_rps": { "median": 0.17, "min": 0.16, "max": 0.18, "n": 3 }, "request_throughput": { "median": 0.35, "min": 0.32, "max": 0.36, "n": 3 }, "completed": [ 60, 60, 60 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1440.0, "max": 1507.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.361 }, "ttft_quantiles_ms": [ 70.6, 105.9, 139.4, 146.2, 149.1, 152.4, 161.3, 162.9, 165.1, 172.6, 181.2, 189.5, 195.6, 204.9, 243.3, 249.4, 258.7, 260.8, 263.7, 270.0, 273.8, 280.4, 283.0, 288.4, 297.2, 313.5, 409.8, 471.2, 734.8, 11751.3, 13244.0, 13325.3, 22973.2, 24754.7, 25146.3, 26342.0, 36689.4, 38790.4, 40004.5, 50497.7, 58990.7, 59709.7, 69144.7, 75203.7, 81445.8, 84389.3, 95331.6, 103083.9, 108082.9, 120421.0, 133481.4 ], "tpot_quantiles_ms": [ 64.0, 67.3, 69.4, 70.5, 70.9, 71.0, 71.4, 72.0, 72.2, 72.6, 73.0, 73.6, 74.2, 75.1, 75.9, 76.7, 77.4, 77.9, 78.4, 80.1, 80.3, 81.1, 81.7, 82.7, 83.0, 86.4, 109.2, 109.8, 111.3, 112.0, 113.2, 114.1, 115.3, 119.2, 123.1, 130.5, 154.5, 165.9, 173.3, 183.0, 184.4, 189.4, 190.8, 191.6, 195.8, 198.0, 199.9, 204.3, 208.5, 211.0, 236.7 ] }, { "x": 4.0, "output_throughput": { "median": 42.51, "min": 38.11, "max": 42.64, "n": 3 }, "median_ttft_ms": { "median": 526.53, "min": 459.41, "max": 1285.24, "n": 3 }, "p99_ttft_ms": { "median": 313593.51, "min": 311992.0, "max": 347318.44, "n": 3 }, "median_tpot_ms": { "median": 123.1, "min": 121.08, "max": 126.17, "n": 3 }, "p99_tpot_ms": { "median": 219.98, "min": 214.62, "max": 240.34, "n": 3 }, "p99_itl_ms": { "median": 416.98, "min": 408.01, "max": 596.72, "n": 3 }, "median_itl_ms": { "median": 113.94, "min": 92.02, "max": 130.65, "n": 3 }, "goodput_rps": { "median": 0.04, "min": 0.04, "max": 0.04, "n": 3 }, "request_throughput": { "median": 0.33, "min": 0.3, "max": 0.33, "n": 3 }, "completed": [ 120, 120, 120 ], "gpu": { "median_sm_mhz": { "median": 1357.5, "min": 1335.0, "max": 1387.5, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.436 }, "ttft_quantiles_ms": [ 72.4, 250.3, 260.1, 263.5, 269.6, 275.4, 281.1, 284.0, 288.7, 291.6, 297.8, 300.9, 309.1, 312.8, 317.3, 319.9, 322.8, 331.2, 335.0, 344.4, 352.2, 360.4, 379.2, 425.2, 473.3, 710.7, 1044.1, 13456.5, 14617.1, 27555.4, 29037.7, 46088.7, 60868.2, 69681.7, 89228.6, 98046.9, 115291.1, 135261.3, 152086.7, 163406.5, 184400.0, 188780.0, 212518.0, 236081.4, 238758.5, 262267.3, 266286.6, 288090.9, 309016.6, 314009.2, 348412.5 ], "tpot_quantiles_ms": [ 68.5, 75.5, 80.4, 85.2, 90.2, 93.9, 98.7, 103.2, 105.4, 108.9, 110.5, 111.2, 112.0, 112.7, 114.0, 115.4, 116.6, 118.0, 119.1, 119.6, 119.9, 120.8, 121.3, 122.0, 122.7, 123.7, 124.4, 124.9, 127.2, 130.3, 155.8, 168.3, 173.6, 177.8, 181.3, 184.8, 189.1, 190.8, 193.5, 196.5, 201.4, 202.5, 205.3, 206.9, 209.4, 211.0, 212.4, 217.3, 222.7, 227.8, 243.1 ] }, { "x": 8.0, "output_throughput": { "median": 41.89, "min": 36.97, "max": 42.31, "n": 3 }, "median_ttft_ms": { "median": 20472.34, "min": 18084.61, "max": 23756.34, "n": 3 }, "p99_ttft_ms": { "median": 684494.02, "min": 677007.87, "max": 774633.37, "n": 3 }, "median_tpot_ms": { "median": 126.84, "min": 126.83, "max": 126.98, "n": 3 }, "p99_tpot_ms": { "median": 219.94, "min": 213.71, "max": 233.93, "n": 3 }, "p99_itl_ms": { "median": 637.32, "min": 402.17, "max": 1000.52, "n": 3 }, "median_itl_ms": { "median": 117.61, "min": 98.5, "max": 135.48, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.33, "min": 0.29, "max": 0.33, "n": 3 }, "completed": [ 240, 240, 240 ], "gpu": { "median_sm_mhz": { "median": 1365.0, "min": 1305.0, "max": 1380.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.465 }, "ttft_quantiles_ms": [ 161.8, 277.6, 332.1, 385.4, 427.4, 483.3, 530.9, 607.1, 749.6, 6748.0, 7263.7, 7628.1, 7954.7, 8249.9, 8648.0, 8919.0, 9457.6, 13957.3, 14481.4, 14857.1, 15044.3, 15303.2, 15986.7, 16482.7, 17003.4, 20299.7, 24111.1, 28683.5, 44492.6, 69665.7, 93813.6, 133562.1, 159482.4, 185438.2, 232505.9, 262236.8, 291524.6, 320853.7, 354567.8, 384571.9, 414087.7, 439810.7, 468406.8, 510569.8, 535589.0, 565940.5, 605208.1, 630578.0, 659900.6, 688875.2, 776136.7 ], "tpot_quantiles_ms": [ 67.8, 71.7, 93.1, 99.6, 100.2, 107.8, 109.5, 112.2, 113.7, 115.7, 116.2, 118.2, 118.7, 119.1, 119.2, 119.9, 121.1, 121.8, 122.7, 123.8, 124.3, 125.1, 125.8, 126.0, 126.5, 126.9, 127.3, 128.0, 145.8, 166.4, 167.9, 169.8, 172.4, 175.0, 176.8, 178.5, 182.8, 184.5, 186.4, 189.0, 191.8, 193.9, 197.1, 201.0, 202.1, 203.2, 204.7, 209.4, 217.7, 222.3, 250.5 ] }, { "x": 16.0, "output_throughput": { "median": 40.25, "min": 34.77, "max": 40.86, "n": 3 }, "median_ttft_ms": { "median": 67489.89, "min": 66427.39, "max": 72552.3, "n": 3 }, "p99_ttft_ms": { "median": 1449287.43, "min": 1426506.82, "max": 1678143.99, "n": 3 }, "median_tpot_ms": { "median": 123.16, "min": 117.78, "max": 132.91, "n": 3 }, "p99_tpot_ms": { "median": 213.44, "min": 210.97, "max": 246.4, "n": 3 }, "p99_itl_ms": { "median": 253.35, "min": 250.26, "max": 254.9, "n": 3 }, "median_itl_ms": { "median": 86.59, "min": 86.01, "max": 86.69, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.31, "min": 0.27, "max": 0.32, "n": 3 }, "completed": [ 480, 480, 480 ], "gpu": { "median_sm_mhz": { "median": 1230.0, "min": 1110.0, "max": 1275.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.492 }, "ttft_quantiles_ms": [ 160.6, 888.4, 1217.5, 1599.9, 2068.6, 12547.4, 12991.7, 13357.7, 13947.6, 23834.1, 24426.8, 24941.7, 25515.4, 34603.2, 35544.8, 36440.8, 36982.3, 45861.1, 46802.8, 47808.9, 48388.8, 57100.3, 59374.1, 59893.2, 60343.0, 67449.8, 72221.1, 78442.3, 137410.7, 196343.8, 263582.7, 336506.5, 396095.5, 457682.1, 516724.4, 581096.9, 650422.4, 712645.3, 777501.3, 838018.0, 895951.6, 960731.4, 1037616.5, 1096700.5, 1158821.1, 1220467.1, 1274415.9, 1346570.9, 1423191.9, 1495756.2, 1710454.8 ], "tpot_quantiles_ms": [ 76.6, 83.7, 84.0, 85.5, 87.7, 88.0, 88.3, 88.6, 91.5, 95.0, 97.2, 97.4, 97.8, 100.5, 104.2, 106.2, 106.5, 106.6, 106.8, 108.3, 114.8, 115.5, 115.8, 116.5, 117.8, 123.1, 132.9, 158.8, 160.6, 162.6, 164.6, 167.9, 170.5, 172.0, 174.1, 177.3, 180.5, 183.0, 185.7, 189.4, 193.0, 196.1, 198.0, 198.8, 200.0, 201.1, 207.0, 211.9, 223.1, 225.9, 257.6 ] } ] }, "sglang-h2h": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 70.64, "min": 67.19, "max": 92.3, "n": 3 }, "median_ttft_ms": { "median": 284.01, "min": 277.57, "max": 285.09, "n": 3 }, "p99_ttft_ms": { "median": 408.4, "min": 404.97, "max": 471.92, "n": 3 }, "median_tpot_ms": { "median": 74.72, "min": 72.12, "max": 76.33, "n": 3 }, "p99_tpot_ms": { "median": 80.97, "min": 80.19, "max": 83.31, "n": 3 }, "p99_itl_ms": { "median": 235.01, "min": 233.87, "max": 235.1, "n": 3 }, "median_itl_ms": { "median": 65.93, "min": 65.52, "max": 66.87, "n": 3 }, "goodput_rps": { "median": 0.55, "min": 0.52, "max": 0.72, "n": 3 }, "request_throughput": { "median": 0.55, "min": 0.52, "max": 0.72, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1485.0, "max": 1485.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 70.8, 71.7, 235.6, 248.1, 249.5, 250.5, 252.3, 252.5, 256.8, 257.2, 258.5, 259.7, 262.3, 263.3, 266.4, 266.8, 270.0, 270.0, 271.4, 272.5, 273.3, 274.1, 274.8, 277.5, 278.5, 282.2, 283.3, 284.7, 285.0, 287.0, 287.6, 288.8, 289.8, 291.0, 293.2, 297.5, 302.6, 305.0, 305.2, 305.5, 306.6, 309.2, 312.8, 315.5, 335.8, 344.7, 370.6, 394.9, 412.8, 463.2, 475.4 ], "tpot_quantiles_ms": [ 60.4, 63.4, 65.4, 66.0, 66.5, 66.7, 67.3, 68.0, 68.1, 68.3, 68.6, 68.8, 69.4, 69.6, 69.8, 71.0, 71.2, 71.5, 71.9, 72.3, 72.8, 72.9, 73.2, 73.4, 73.7, 74.1, 74.7, 74.8, 74.9, 75.4, 76.0, 76.2, 76.2, 76.3, 76.4, 76.5, 77.0, 77.5, 77.7, 77.7, 77.8, 78.6, 78.9, 79.1, 79.1, 79.4, 80.1, 80.2, 80.3, 81.6, 83.7 ] }, { "x": 1.0, "output_throughput": { "median": 123.31, "min": 118.39, "max": 153.57, "n": 3 }, "median_ttft_ms": { "median": 165.75, "min": 161.51, "max": 174.46, "n": 3 }, "p99_ttft_ms": { "median": 219.71, "min": 204.21, "max": 222.27, "n": 3 }, "median_tpot_ms": { "median": 74.01, "min": 72.46, "max": 74.59, "n": 3 }, "p99_tpot_ms": { "median": 76.31, "min": 76.16, "max": 77.77, "n": 3 }, "p99_itl_ms": { "median": 113.88, "min": 112.9, "max": 165.71, "n": 3 }, "median_itl_ms": { "median": 68.78, "min": 67.29, "max": 69.12, "n": 3 }, "goodput_rps": { "median": 0.96, "min": 0.92, "max": 1.2, "n": 3 }, "request_throughput": { "median": 0.96, "min": 0.92, "max": 1.2, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1500.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 71.3, 71.4, 117.3, 131.2, 135.8, 137.5, 139.3, 140.0, 141.7, 142.0, 143.1, 143.6, 145.2, 146.6, 147.1, 149.9, 154.5, 157.3, 157.8, 158.7, 161.1, 161.3, 162.9, 164.9, 165.7, 166.2, 168.2, 170.8, 173.2, 175.8, 178.2, 178.5, 179.7, 182.8, 185.2, 186.4, 186.5, 187.9, 188.3, 190.6, 191.6, 191.8, 193.6, 195.8, 199.7, 201.3, 202.2, 204.7, 208.7, 215.7, 225.2 ], "tpot_quantiles_ms": [ 62.5, 65.3, 65.9, 67.3, 67.7, 68.2, 68.6, 69.0, 69.2, 69.6, 70.0, 70.1, 70.2, 70.4, 70.8, 71.1, 71.2, 71.6, 72.3, 72.5, 72.6, 72.7, 72.8, 73.2, 73.5, 73.5, 73.8, 73.8, 74.0, 74.0, 74.2, 74.2, 74.3, 74.5, 74.6, 74.9, 75.0, 75.2, 75.5, 75.6, 76.1, 76.1, 76.1, 76.2, 76.3, 76.3, 76.6, 76.7, 77.0, 77.4, 77.8 ] }, { "x": 2.0, "output_throughput": { "median": 226.29, "min": 215.51, "max": 230.07, "n": 3 }, "median_ttft_ms": { "median": 220.6, "min": 211.47, "max": 236.29, "n": 3 }, "p99_ttft_ms": { "median": 603.21, "min": 539.35, "max": 1428.31, "n": 3 }, "median_tpot_ms": { "median": 99.9, "min": 92.89, "max": 105.89, "n": 3 }, "p99_tpot_ms": { "median": 120.38, "min": 116.71, "max": 122.07, "n": 3 }, "p99_itl_ms": { "median": 386.62, "min": 383.92, "max": 429.0, "n": 3 }, "median_itl_ms": { "median": 80.39, "min": 77.99, "max": 82.91, "n": 3 }, "goodput_rps": { "median": 0.88, "min": 0.54, "max": 1.18, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.68, "max": 1.8, "n": 3 }, "completed": [ 60, 60, 60 ], "gpu": { "median_sm_mhz": { "median": 1432.5, "min": 1410.0, "max": 1500.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 70.9, 112.4, 130.0, 138.3, 144.7, 147.2, 151.6, 156.6, 158.0, 161.2, 163.8, 168.7, 171.3, 173.1, 177.9, 180.6, 184.4, 190.4, 192.5, 198.8, 200.7, 202.6, 203.5, 207.1, 213.0, 224.7, 239.4, 251.5, 278.2, 288.0, 292.5, 297.7, 312.7, 315.7, 319.2, 321.0, 327.7, 329.9, 345.7, 354.3, 376.4, 388.4, 401.9, 415.8, 427.7, 439.6, 459.4, 529.9, 598.9, 1170.3, 1621.4 ], "tpot_quantiles_ms": [ 71.0, 76.4, 78.6, 80.9, 82.5, 83.5, 84.1, 84.3, 84.6, 85.6, 85.8, 86.0, 86.0, 88.4, 89.3, 89.8, 90.4, 91.5, 92.7, 93.8, 94.5, 94.9, 95.6, 97.0, 97.7, 99.2, 100.4, 100.6, 101.9, 102.3, 103.8, 105.1, 106.5, 107.5, 108.6, 109.1, 110.2, 111.4, 113.0, 114.1, 115.6, 116.2, 116.9, 117.5, 118.8, 119.5, 119.7, 120.0, 120.3, 121.4, 122.1 ] }, { "x": 4.0, "output_throughput": { "median": 257.46, "min": 254.93, "max": 259.72, "n": 3 }, "median_ttft_ms": { "median": 9575.25, "min": 8423.14, "max": 9843.31, "n": 3 }, "p99_ttft_ms": { "median": 24244.88, "min": 21501.98, "max": 25364.5, "n": 3 }, "median_tpot_ms": { "median": 115.08, "min": 113.66, "max": 116.93, "n": 3 }, "p99_tpot_ms": { "median": 127.23, "min": 126.58, "max": 128.25, "n": 3 }, "p99_itl_ms": { "median": 680.11, "min": 383.16, "max": 810.9, "n": 3 }, "median_itl_ms": { "median": 83.71, "min": 83.66, "max": 83.78, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.01, "min": 1.99, "max": 2.03, "n": 3 }, "completed": [ 120, 120, 120 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1320.0, "max": 1350.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 71.2, 274.5, 336.4, 389.9, 416.9, 524.8, 593.0, 645.9, 713.8, 761.6, 812.9, 865.3, 928.0, 1032.2, 1262.1, 1648.4, 7002.1, 7564.3, 7716.3, 7864.4, 8129.6, 8424.1, 8551.8, 8859.3, 8999.3, 9183.2, 9421.5, 9809.0, 10038.5, 10174.1, 10515.1, 12864.8, 14244.4, 14554.4, 14981.9, 15334.3, 15775.0, 16156.4, 16260.5, 16435.0, 16546.7, 16794.5, 16980.4, 17297.0, 17607.1, 18518.4, 20058.0, 21113.0, 23818.7, 24206.8, 25429.4 ], "tpot_quantiles_ms": [ 66.4, 68.5, 72.7, 76.4, 93.2, 95.8, 99.1, 100.3, 100.8, 101.2, 101.6, 106.4, 108.8, 109.8, 110.7, 110.7, 111.2, 113.3, 113.3, 113.8, 114.0, 114.7, 115.0, 115.1, 115.3, 115.5, 116.2, 116.9, 118.4, 119.0, 119.3, 120.1, 120.1, 120.7, 121.3, 121.9, 122.4, 123.4, 123.7, 124.1, 124.6, 125.1, 125.3, 125.7, 125.9, 126.3, 126.7, 126.8, 127.1, 127.9, 128.3 ] }, { "x": 8.0, "output_throughput": { "median": 253.87, "min": 250.66, "max": 260.67, "n": 3 }, "median_ttft_ms": { "median": 37210.89, "min": 35532.61, "max": 37345.35, "n": 3 }, "p99_ttft_ms": { "median": 82070.83, "min": 81045.76, "max": 84231.25, "n": 3 }, "median_tpot_ms": { "median": 105.18, "min": 104.57, "max": 105.34, "n": 3 }, "p99_tpot_ms": { "median": 151.24, "min": 149.8, "max": 152.97, "n": 3 }, "p99_itl_ms": { "median": 277.35, "min": 258.92, "max": 316.6, "n": 3 }, "median_itl_ms": { "median": 83.71, "min": 83.7, "max": 83.8, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.98, "min": 1.96, "max": 2.04, "n": 3 }, "completed": [ 240, 240, 240 ], "gpu": { "median_sm_mhz": { "median": 1215.0, "min": 1192.5, "max": 1230.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.322 }, "ttft_quantiles_ms": [ 70.5, 724.5, 1045.4, 1548.2, 1747.8, 1931.0, 2156.3, 2598.5, 12334.2, 13090.6, 13469.4, 13740.5, 13896.5, 14196.9, 14430.4, 14578.8, 24567.7, 24921.1, 25236.6, 25524.2, 25746.3, 25979.2, 26257.3, 26706.0, 35626.3, 36006.5, 36471.2, 37343.8, 37547.9, 38010.9, 38283.0, 46752.2, 48282.4, 48738.5, 49324.0, 49679.4, 50075.6, 51330.2, 52676.2, 60633.6, 63057.3, 63645.1, 64556.4, 65496.1, 66100.9, 67637.4, 68758.7, 80080.4, 81061.8, 82347.5, 84592.7 ], "tpot_quantiles_ms": [ 76.9, 82.0, 85.3, 85.4, 85.7, 86.5, 86.8, 88.1, 88.7, 89.0, 89.9, 90.5, 90.8, 93.0, 94.5, 95.4, 97.2, 98.0, 98.2, 98.3, 99.2, 99.7, 101.6, 103.6, 104.0, 105.2, 106.0, 107.1, 107.6, 108.6, 108.7, 108.8, 111.3, 113.8, 114.3, 116.3, 116.6, 116.9, 117.9, 118.1, 118.7, 120.2, 122.4, 123.9, 125.7, 129.8, 130.1, 133.8, 139.3, 151.2, 175.5 ] }, { "x": 16.0, "output_throughput": { "median": 218.24, "min": 215.57, "max": 228.96, "n": 3 }, "median_ttft_ms": { "median": 97542.26, "min": 94794.53, "max": 100312.78, "n": 3 }, "p99_ttft_ms": { "median": 238041.74, "min": 226583.4, "max": 242851.07, "n": 3 }, "median_tpot_ms": { "median": 115.2, "min": 109.94, "max": 115.88, "n": 3 }, "p99_tpot_ms": { "median": 188.69, "min": 182.16, "max": 194.29, "n": 3 }, "p99_itl_ms": { "median": 137.39, "min": 119.78, "max": 163.36, "n": 3 }, "median_itl_ms": { "median": 84.57, "min": 84.44, "max": 84.74, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.71, "min": 1.68, "max": 1.79, "n": 3 }, "completed": [ 480, 480, 480 ], "gpu": { "median_sm_mhz": { "median": 600.0, "min": 570.0, "max": 765.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.676 }, "ttft_quantiles_ms": [ 71.3, 1579.5, 2610.5, 3537.2, 14917.1, 15905.5, 16834.8, 17653.9, 29245.8, 30147.7, 30900.6, 31617.1, 43139.6, 44097.6, 44948.1, 45857.5, 57597.1, 58629.0, 59906.0, 62164.7, 73792.1, 75654.7, 77921.2, 81592.4, 93747.6, 97211.0, 99418.4, 103630.3, 115285.0, 118885.1, 121268.6, 132669.7, 138467.8, 141223.9, 145194.3, 155449.4, 161095.3, 165371.3, 168870.4, 179111.0, 184337.1, 188961.7, 193349.4, 201248.7, 208006.6, 214525.8, 218388.4, 224756.0, 233500.9, 238495.2, 243212.4 ], "tpot_quantiles_ms": [ 83.7, 84.0, 84.2, 87.6, 88.5, 88.6, 89.3, 91.6, 93.1, 96.7, 97.3, 97.9, 98.2, 98.7, 99.6, 102.0, 103.2, 105.5, 106.5, 107.3, 107.5, 108.3, 109.9, 110.2, 111.7, 114.5, 116.5, 116.8, 117.3, 118.5, 120.4, 122.3, 124.0, 127.1, 128.7, 131.0, 132.7, 136.2, 138.8, 141.5, 145.9, 147.5, 152.1, 153.9, 159.6, 161.9, 165.8, 167.7, 173.3, 176.7, 204.1 ] } ] }, "vllm": { "kv_pool_tokens": 24096, "points": [ { "x": 0.5, "output_throughput": { "median": 70.58, "min": 67.22, "max": 92.33, "n": 3 }, "median_ttft_ms": { "median": 317.04, "min": 311.89, "max": 330.5, "n": 3 }, "p99_ttft_ms": { "median": 474.68, "min": 410.52, "max": 494.51, "n": 3 }, "median_tpot_ms": { "median": 72.71, "min": 70.41, "max": 74.05, "n": 3 }, "p99_tpot_ms": { "median": 78.16, "min": 76.27, "max": 79.74, "n": 3 }, "p99_itl_ms": { "median": 214.09, "min": 208.17, "max": 214.82, "n": 3 }, "median_itl_ms": { "median": 65.58, "min": 65.37, "max": 66.14, "n": 3 }, "goodput_rps": { "median": 0.55, "min": 0.53, "max": 0.72, "n": 3 }, "request_throughput": { "median": 0.55, "min": 0.53, "max": 0.72, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 76.9, 77.2, 262.4, 268.5, 273.8, 275.2, 278.3, 279.6, 285.5, 288.2, 289.1, 289.4, 292.1, 294.4, 298.3, 301.5, 303.9, 304.7, 306.3, 307.6, 310.3, 311.9, 312.5, 313.5, 315.1, 317.3, 319.0, 320.3, 324.6, 324.7, 327.2, 327.3, 329.5, 331.5, 331.6, 334.7, 336.5, 339.5, 342.2, 345.2, 346.8, 347.5, 360.3, 379.0, 389.5, 398.6, 411.2, 427.8, 468.4, 475.4, 503.1 ], "tpot_quantiles_ms": [ 60.6, 62.9, 65.1, 65.5, 65.8, 66.0, 66.6, 67.1, 67.3, 67.5, 67.7, 67.8, 68.4, 68.5, 68.7, 69.7, 69.8, 69.9, 70.1, 70.5, 71.0, 71.0, 71.4, 71.7, 72.0, 72.2, 72.6, 72.8, 73.0, 73.1, 73.4, 73.7, 73.9, 74.1, 74.2, 74.2, 74.5, 74.6, 75.1, 75.1, 75.2, 75.2, 75.5, 75.9, 76.1, 76.2, 76.3, 76.7, 77.7, 78.5, 80.1 ] }, { "x": 1.0, "output_throughput": { "median": 123.6, "min": 118.36, "max": 154.52, "n": 3 }, "median_ttft_ms": { "median": 179.89, "min": 179.31, "max": 182.03, "n": 3 }, "p99_ttft_ms": { "median": 212.56, "min": 211.75, "max": 221.72, "n": 3 }, "median_tpot_ms": { "median": 68.16, "min": 67.04, "max": 68.61, "n": 3 }, "p99_tpot_ms": { "median": 68.76, "min": 68.42, "max": 69.54, "n": 3 }, "p99_itl_ms": { "median": 77.41, "min": 77.32, "max": 79.09, "n": 3 }, "median_itl_ms": { "median": 67.24, "min": 66.42, "max": 67.72, "n": 3 }, "goodput_rps": { "median": 0.97, "min": 0.92, "max": 1.21, "n": 3 }, "request_throughput": { "median": 0.97, "min": 0.92, "max": 1.21, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 75.5, 77.0, 146.2, 149.0, 153.9, 154.3, 154.6, 158.1, 159.7, 160.5, 161.9, 162.6, 164.1, 166.2, 167.3, 167.9, 169.9, 170.5, 171.0, 172.2, 174.8, 175.0, 175.8, 177.2, 178.1, 180.5, 182.2, 183.2, 184.3, 185.7, 187.7, 188.5, 191.0, 192.1, 192.9, 195.4, 196.6, 199.1, 200.1, 201.8, 202.4, 202.5, 203.6, 204.4, 207.3, 208.6, 209.2, 211.0, 212.3, 213.2, 222.7 ], "tpot_quantiles_ms": [ 62.3, 64.3, 64.7, 65.1, 65.5, 65.5, 65.6, 65.6, 65.9, 66.1, 66.4, 66.4, 66.5, 66.6, 66.8, 67.0, 67.1, 67.2, 67.4, 67.6, 67.7, 67.7, 67.9, 68.0, 68.0, 68.0, 68.1, 68.1, 68.1, 68.2, 68.2, 68.2, 68.2, 68.3, 68.3, 68.4, 68.4, 68.4, 68.5, 68.6, 68.6, 68.7, 68.7, 68.8, 68.9, 69.0, 69.1, 69.2, 69.3, 69.5, 69.5 ] }, { "x": 2.0, "output_throughput": { "median": 227.82, "min": 216.47, "max": 231.82, "n": 3 }, "median_ttft_ms": { "median": 232.88, "min": 222.03, "max": 249.53, "n": 3 }, "p99_ttft_ms": { "median": 722.35, "min": 673.98, "max": 755.14, "n": 3 }, "median_tpot_ms": { "median": 88.7, "min": 84.46, "max": 93.39, "n": 3 }, "p99_tpot_ms": { "median": 109.06, "min": 107.28, "max": 111.6, "n": 3 }, "p99_itl_ms": { "median": 354.8, "min": 351.04, "max": 357.97, "n": 3 }, "median_itl_ms": { "median": 78.08, "min": 76.21, "max": 79.66, "n": 3 }, "goodput_rps": { "median": 1.36, "min": 1.09, "max": 1.44, "n": 3 }, "request_throughput": { "median": 1.78, "min": 1.69, "max": 1.81, "n": 3 }, "completed": [ 60, 60, 60 ], "gpu": { "median_sm_mhz": { "median": 1417.5, "min": 1395.0, "max": 1485.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 74.8, 146.8, 160.1, 166.9, 169.7, 170.2, 174.1, 176.4, 176.7, 179.7, 182.9, 185.6, 188.4, 192.5, 197.6, 199.9, 205.3, 206.9, 208.3, 211.2, 214.7, 219.1, 220.7, 221.7, 223.8, 242.9, 250.6, 296.8, 324.0, 327.8, 331.3, 340.1, 351.4, 357.7, 375.9, 384.1, 388.6, 398.9, 414.4, 447.3, 463.0, 481.1, 486.2, 507.2, 541.3, 571.2, 581.8, 612.8, 621.7, 706.8, 775.8 ], "tpot_quantiles_ms": [ 69.0, 71.4, 72.3, 73.6, 73.7, 74.0, 74.4, 75.6, 75.9, 76.0, 76.5, 77.4, 78.3, 79.3, 80.2, 80.7, 81.9, 82.1, 82.6, 84.4, 84.6, 85.5, 86.3, 86.7, 87.3, 88.6, 89.5, 90.8, 91.2, 92.4, 92.7, 93.3, 94.4, 95.5, 96.7, 97.8, 98.8, 99.9, 101.3, 102.5, 103.5, 104.2, 105.2, 106.1, 106.5, 107.2, 107.3, 107.7, 109.0, 111.4, 111.6 ] }, { "x": 4.0, "output_throughput": { "median": 286.62, "min": 283.67, "max": 291.74, "n": 3 }, "median_ttft_ms": { "median": 8303.55, "min": 7744.05, "max": 9672.44, "n": 3 }, "p99_ttft_ms": { "median": 18006.79, "min": 16346.78, "max": 19649.84, "n": 3 }, "median_tpot_ms": { "median": 126.9, "min": 126.61, "max": 128.43, "n": 3 }, "p99_tpot_ms": { "median": 186.97, "min": 174.48, "max": 190.9, "n": 3 }, "p99_itl_ms": { "median": 621.28, "min": 620.04, "max": 627.17, "n": 3 }, "median_itl_ms": { "median": 83.45, "min": 83.38, "max": 83.55, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.24, "min": 2.22, "max": 2.28, "n": 3 }, "completed": [ 120, 120, 120 ], "gpu": { "median_sm_mhz": { "median": 1245.0, "min": 1200.0, "max": 1290.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 76.5, 306.7, 379.0, 434.7, 469.4, 555.4, 616.4, 687.0, 790.3, 831.8, 881.4, 959.8, 988.3, 1029.9, 1089.8, 1143.7, 1268.7, 1517.6, 1865.1, 7326.4, 7575.6, 7712.0, 7828.4, 7967.8, 8189.5, 8316.8, 8477.6, 9150.5, 9474.3, 9643.7, 9734.6, 9857.6, 10049.2, 10361.2, 11195.6, 12517.2, 13611.9, 14143.5, 14575.3, 15259.0, 15532.3, 15792.4, 16112.8, 16321.5, 16719.6, 17257.1, 17419.6, 17547.1, 17883.0, 18505.7, 19854.3 ], "tpot_quantiles_ms": [ 82.3, 84.4, 87.2, 89.4, 90.4, 94.8, 98.1, 101.7, 102.8, 106.1, 107.8, 110.4, 113.9, 116.2, 119.0, 122.3, 123.1, 124.1, 124.8, 125.3, 125.5, 126.1, 126.6, 127.0, 127.0, 127.6, 127.9, 128.0, 128.1, 128.4, 128.4, 129.1, 129.4, 129.6, 129.6, 130.0, 131.1, 131.4, 131.6, 132.0, 132.5, 132.9, 133.1, 133.9, 134.8, 135.8, 136.9, 138.0, 158.6, 176.6, 202.2 ] }, { "x": 8.0, "output_throughput": { "median": 261.83, "min": 259.38, "max": 264.28, "n": 3 }, "median_ttft_ms": { "median": 27510.33, "min": 26282.57, "max": 27958.75, "n": 3 }, "p99_ttft_ms": { "median": 78008.36, "min": 75396.54, "max": 80381.07, "n": 3 }, "median_tpot_ms": { "median": 125.97, "min": 125.65, "max": 126.15, "n": 3 }, "p99_tpot_ms": { "median": 248.32, "min": 242.27, "max": 249.96, "n": 3 }, "p99_itl_ms": { "median": 1183.77, "min": 1088.72, "max": 1360.91, "n": 3 }, "median_itl_ms": { "median": 83.83, "min": 83.51, "max": 84.01, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.05, "min": 2.03, "max": 2.06, "n": 3 }, "completed": [ 240, 240, 240 ], "gpu": { "median_sm_mhz": { "median": 1035.0, "min": 1005.0, "max": 1035.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.38 }, "ttft_quantiles_ms": [ 76.2, 734.9, 1214.9, 1657.9, 1852.0, 2007.0, 2230.7, 2393.1, 2748.8, 3497.8, 13738.5, 14529.5, 14707.7, 14872.7, 15045.1, 15161.3, 15362.9, 16243.5, 25747.4, 26034.3, 26305.2, 26760.6, 26996.3, 27178.0, 27424.5, 27716.9, 28069.1, 37717.1, 38703.3, 39244.6, 39590.8, 39841.1, 40382.9, 40865.4, 42216.8, 51577.1, 52568.3, 53458.3, 54070.8, 54852.4, 55428.7, 56483.5, 57292.5, 59544.7, 70841.3, 72793.6, 74004.3, 74968.1, 76577.0, 78096.6, 80709.3 ], "tpot_quantiles_ms": [ 85.8, 94.1, 108.0, 119.7, 119.7, 119.9, 120.4, 120.5, 121.8, 122.0, 122.2, 122.3, 122.4, 122.5, 122.8, 123.3, 123.5, 123.8, 124.0, 124.3, 124.8, 125.0, 125.2, 125.3, 125.5, 125.9, 126.9, 127.3, 128.2, 129.9, 131.0, 133.7, 137.1, 139.3, 141.6, 143.3, 148.9, 154.1, 160.4, 162.9, 164.7, 167.1, 169.2, 171.2, 174.3, 177.4, 180.5, 186.9, 206.2, 231.7, 274.8 ] }, { "x": 16.0, "output_throughput": { "median": 205.77, "min": 204.75, "max": 208.41, "n": 3 }, "median_ttft_ms": { "median": 94322.33, "min": 91404.02, "max": 94402.18, "n": 3 }, "p99_ttft_ms": { "median": 256061.1, "min": 249556.02, "max": 256733.32, "n": 3 }, "median_tpot_ms": { "median": 185.15, "min": 182.17, "max": 190.82, "n": 3 }, "p99_tpot_ms": { "median": 321.75, "min": 317.28, "max": 330.38, "n": 3 }, "p99_itl_ms": { "median": 1448.33, "min": 1448.2, "max": 1483.43, "n": 3 }, "median_itl_ms": { "median": 92.5, "min": 92.41, "max": 104.11, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.61, "min": 1.6, "max": 1.63, "n": 3 }, "completed": [ 480, 480, 480 ], "gpu": { "median_sm_mhz": { "median": 450.0, "min": 450.0, "max": 465.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.721 }, "ttft_quantiles_ms": [ 76.9, 1610.9, 2545.0, 3482.4, 4376.2, 16505.6, 17515.6, 18447.0, 19460.6, 31104.9, 31975.3, 32713.2, 33451.6, 34772.6, 46655.1, 47779.7, 48801.0, 50348.2, 63085.3, 65170.2, 66854.4, 69429.3, 84658.9, 88073.2, 90671.4, 93347.0, 108123.3, 112101.2, 114975.3, 117247.8, 121420.9, 138094.4, 141296.9, 143387.9, 147634.6, 164914.2, 168098.6, 170978.8, 174153.2, 190497.3, 195587.1, 198004.8, 201842.5, 216652.8, 222954.1, 226054.7, 228490.7, 232261.7, 250261.5, 254069.6, 257728.6 ], "tpot_quantiles_ms": [ 102.8, 119.3, 120.6, 120.9, 121.1, 122.3, 122.7, 123.0, 123.7, 124.0, 124.1, 124.3, 126.2, 127.7, 130.5, 136.6, 142.2, 146.9, 155.3, 163.9, 169.0, 174.1, 179.7, 182.2, 183.9, 186.3, 186.6, 190.8, 192.7, 198.2, 200.1, 201.7, 202.1, 202.8, 203.9, 205.3, 206.6, 207.1, 208.2, 208.4, 209.0, 209.6, 210.9, 211.4, 212.4, 213.1, 215.4, 247.2, 275.7, 308.4, 347.1 ] } ] } }, "quantiles": [ 0.0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12, 0.14, 0.16, 0.18, 0.2, 0.22, 0.24, 0.26, 0.28, 0.3, 0.32, 0.34, 0.36, 0.38, 0.4, 0.42, 0.44, 0.46, 0.48, 0.5, 0.52, 0.54, 0.56, 0.58, 0.6, 0.62, 0.64, 0.66, 0.68, 0.7, 0.72, 0.74, 0.76, 0.78, 0.8, 0.82, 0.84, 0.86, 0.88, 0.9, 0.92, 0.94, 0.96, 0.98, 1.0 ], "prefix": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 113.56, "min": 113.25, "max": 113.56, "n": 3 }, "median_ttft_ms": { "median": 349.39, "min": 320.73, "max": 351.95, "n": 3 }, "p99_ttft_ms": { "median": 486.41, "min": 474.1, "max": 495.91, "n": 3 }, "median_tpot_ms": { "median": 66.12, "min": 66.12, "max": 66.63, "n": 3 }, "p99_tpot_ms": { "median": 70.11, "min": 70.06, "max": 70.18, "n": 3 }, "p99_itl_ms": { "median": 149.5, "min": 71.59, "max": 182.52, "n": 3 }, "median_itl_ms": { "median": 64.14, "min": 64.13, "max": 64.15, "n": 3 }, "goodput_rps": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 84.4, 105.7, 109.5, 109.8, 110.5, 111.2, 113.1, 289.0, 294.0, 310.2, 316.7, 318.3, 318.9, 320.0, 335.8, 337.4, 344.1, 345.7, 346.3, 347.0, 347.5, 348.1, 348.8, 348.9, 349.4, 350.0, 350.6, 351.1, 351.4, 352.3, 353.0, 354.5, 355.7, 356.7, 357.5, 358.2, 408.1, 462.2, 462.8, 463.8, 464.9, 467.7, 469.1, 471.0, 472.1, 472.5, 472.9, 473.3, 474.6, 484.9, 537.1 ], "tpot_quantiles_ms": [ 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 65.4, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.6, 66.7, 66.7, 66.7, 66.7, 67.1, 67.3, 68.4, 69.8, 69.9, 70.0, 70.0, 70.1, 70.2 ] }, { "x": 256.0, "output_throughput": { "median": 110.83, "min": 110.59, "max": 111.06, "n": 3 }, "median_ttft_ms": { "median": 474.02, "min": 330.03, "max": 474.18, "n": 3 }, "p99_ttft_ms": { "median": 770.57, "min": 690.15, "max": 839.11, "n": 3 }, "median_tpot_ms": { "median": 67.26, "min": 67.17, "max": 67.31, "n": 3 }, "p99_tpot_ms": { "median": 73.15, "min": 72.66, "max": 73.62, "n": 3 }, "p99_itl_ms": { "median": 75.65, "min": 73.55, "max": 174.94, "n": 3 }, "median_itl_ms": { "median": 64.74, "min": 64.73, "max": 64.75, "n": 3 }, "goodput_rps": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "request_throughput": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1455.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 91.5, 103.9, 112.5, 113.5, 113.9, 115.7, 118.2, 255.8, 284.8, 289.5, 290.6, 291.5, 292.1, 292.7, 293.8, 294.6, 295.5, 296.1, 310.5, 315.7, 317.9, 324.0, 326.7, 330.2, 467.2, 470.1, 473.4, 474.0, 474.2, 474.9, 475.0, 475.5, 476.2, 478.1, 478.4, 478.9, 479.4, 479.9, 481.2, 482.8, 483.5, 484.4, 485.3, 492.3, 497.9, 634.7, 636.1, 714.1, 780.0, 781.4, 936.6 ], "tpot_quantiles_ms": [ 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 67.2, 67.2, 67.2, 67.3, 67.3, 67.3, 67.3, 67.4, 67.5, 67.5, 67.5, 67.6, 67.6, 67.7, 67.7, 67.8, 67.8, 67.8, 67.9, 67.9, 67.9, 68.5, 70.5, 70.6, 70.6, 70.7, 70.8, 70.9, 78.2 ] }, { "x": 1024.0, "output_throughput": { "median": 105.76, "min": 103.04, "max": 105.95, "n": 3 }, "median_ttft_ms": { "median": 461.82, "min": 458.0, "max": 480.1, "n": 3 }, "p99_ttft_ms": { "median": 1904.82, "min": 1732.6, "max": 1984.87, "n": 3 }, "median_tpot_ms": { "median": 66.71, "min": 66.7, "max": 66.78, "n": 3 }, "p99_tpot_ms": { "median": 89.75, "min": 83.89, "max": 90.29, "n": 3 }, "p99_itl_ms": { "median": 72.69, "min": 70.48, "max": 72.95, "n": 3 }, "median_itl_ms": { "median": 66.41, "min": 66.16, "max": 66.64, "n": 3 }, "goodput_rps": { "median": 1.47, "min": 1.28, "max": 1.52, "n": 3 }, "request_throughput": { "median": 1.65, "min": 1.61, "max": 1.66, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1455.0, "max": 1470.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 92.7, 111.1, 117.8, 119.4, 120.2, 156.1, 159.4, 213.9, 216.1, 219.4, 260.6, 439.8, 441.2, 452.9, 453.7, 455.3, 456.7, 457.5, 457.9, 458.4, 459.1, 459.2, 459.5, 459.8, 460.4, 461.2, 461.5, 461.9, 462.6, 463.8, 464.7, 465.6, 466.6, 467.5, 468.5, 471.9, 473.3, 474.9, 479.7, 481.5, 485.2, 492.9, 497.0, 499.0, 1164.9, 1256.0, 1386.9, 1389.8, 1688.1, 1905.1, 1985.1 ], "tpot_quantiles_ms": [ 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 68.7, 69.7, 70.0, 70.3, 70.6, 70.7, 71.0, 71.5, 72.0, 72.2, 72.3, 77.0, 79.1, 87.3, 96.1 ] }, { "x": 2048.0, "output_throughput": { "median": 100.29, "min": 99.5, "max": 102.42, "n": 3 }, "median_ttft_ms": { "median": 440.47, "min": 439.92, "max": 484.47, "n": 3 }, "p99_ttft_ms": { "median": 2326.02, "min": 2038.87, "max": 2693.02, "n": 3 }, "median_tpot_ms": { "median": 70.56, "min": 68.1, "max": 70.58, "n": 3 }, "p99_tpot_ms": { "median": 98.06, "min": 95.65, "max": 106.65, "n": 3 }, "p99_itl_ms": { "median": 152.27, "min": 74.43, "max": 153.86, "n": 3 }, "median_itl_ms": { "median": 69.1, "min": 68.14, "max": 69.14, "n": 3 }, "goodput_rps": { "median": 1.4, "min": 1.36, "max": 1.48, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.55, "max": 1.6, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1440.0, "max": 1455.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 100.8, 116.4, 124.9, 163.3, 166.0, 166.3, 168.2, 223.0, 421.4, 435.1, 435.6, 436.4, 437.4, 437.6, 438.3, 438.8, 438.9, 439.3, 439.7, 440.0, 440.3, 441.1, 441.4, 442.1, 442.9, 443.8, 474.4, 477.3, 479.1, 481.4, 483.3, 484.2, 484.9, 485.7, 486.5, 488.0, 489.3, 490.9, 491.8, 492.5, 493.6, 494.9, 517.0, 518.9, 522.0, 528.4, 1639.3, 1888.2, 2038.7, 2691.4, 2693.4 ], "tpot_quantiles_ms": [ 68.0, 68.0, 68.0, 68.0, 68.0, 68.0, 68.0, 68.1, 68.1, 68.1, 68.1, 69.1, 69.2, 69.2, 69.2, 69.2, 69.2, 69.2, 69.2, 69.3, 69.3, 69.3, 69.4, 69.4, 69.5, 70.5, 70.5, 70.5, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.7, 70.7, 71.3, 72.9, 73.3, 73.8, 74.3, 74.9, 75.0, 75.1, 76.2, 94.3, 110.2 ] }, { "x": 4096.0, "output_throughput": { "median": 88.87, "min": 88.21, "max": 89.64, "n": 3 }, "median_ttft_ms": { "median": 478.78, "min": 475.51, "max": 531.4, "n": 3 }, "p99_ttft_ms": { "median": 4287.82, "min": 3284.97, "max": 4795.48, "n": 3 }, "median_tpot_ms": { "median": 75.53, "min": 75.51, "max": 77.08, "n": 3 }, "p99_tpot_ms": { "median": 123.69, "min": 109.88, "max": 148.2, "n": 3 }, "p99_itl_ms": { "median": 84.7, "min": 83.66, "max": 159.43, "n": 3 }, "median_itl_ms": { "median": 74.03, "min": 73.92, "max": 74.14, "n": 3 }, "goodput_rps": { "median": 1.21, "min": 1.05, "max": 1.22, "n": 3 }, "request_throughput": { "median": 1.39, "min": 1.38, "max": 1.4, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1440.0, "min": 1440.0, "max": 1455.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 144.8, 168.4, 178.4, 182.6, 216.8, 219.1, 223.1, 313.9, 371.3, 373.2, 374.7, 377.7, 410.5, 412.4, 413.6, 415.5, 417.6, 464.8, 469.2, 472.7, 474.6, 476.4, 477.0, 477.4, 478.0, 478.6, 479.6, 480.8, 482.0, 548.9, 561.2, 582.2, 583.1, 585.1, 586.5, 587.6, 588.2, 588.6, 589.3, 590.5, 592.0, 594.8, 599.4, 615.9, 2041.3, 2043.5, 2685.0, 2730.3, 3285.4, 3955.9, 4852.0 ], "tpot_quantiles_ms": [ 71.8, 71.9, 72.0, 72.1, 72.1, 73.2, 73.2, 73.2, 74.1, 74.1, 74.2, 74.2, 74.2, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.4, 75.4, 75.5, 75.5, 75.5, 75.5, 75.6, 75.6, 75.6, 75.6, 76.4, 76.5, 77.1, 77.1, 77.1, 77.1, 77.2, 77.6, 77.7, 78.0, 78.4, 80.0, 80.2, 80.3, 80.4, 80.5, 84.3, 89.0, 108.7, 109.6, 148.7 ] } ] }, "no-radix-cache": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 113.33, "min": 113.27, "max": 113.35, "n": 3 }, "median_ttft_ms": { "median": 378.82, "min": 341.12, "max": 380.57, "n": 3 }, "p99_ttft_ms": { "median": 480.01, "min": 471.94, "max": 510.8, "n": 3 }, "median_tpot_ms": { "median": 65.53, "min": 64.28, "max": 66.04, "n": 3 }, "p99_tpot_ms": { "median": 70.15, "min": 70.07, "max": 70.31, "n": 3 }, "p99_itl_ms": { "median": 69.2, "min": 68.92, "max": 70.41, "n": 3 }, "median_itl_ms": { "median": 64.15, "min": 64.12, "max": 64.16, "n": 3 }, "goodput_rps": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 100.7, 103.5, 103.8, 104.3, 105.1, 105.5, 106.1, 248.8, 275.8, 278.6, 279.6, 280.4, 281.6, 282.4, 282.9, 283.6, 284.0, 284.4, 285.1, 286.9, 308.1, 311.1, 311.9, 312.9, 318.4, 341.8, 380.4, 381.8, 444.6, 457.2, 460.1, 463.3, 463.9, 464.7, 465.0, 465.9, 467.3, 467.7, 468.2, 468.4, 469.1, 470.2, 471.0, 471.6, 472.0, 472.8, 476.8, 479.3, 480.2, 489.0, 546.4 ], "tpot_quantiles_ms": [ 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 66.0, 66.0, 66.7, 66.8, 66.8, 66.8, 66.8, 67.1, 67.1, 67.1, 67.2, 67.2, 67.2, 67.2, 67.2, 67.2, 67.3, 67.4, 67.5, 68.6, 69.9, 70.0, 70.0, 70.1, 70.1, 70.4 ] }, { "x": 256.0, "output_throughput": { "median": 99.38, "min": 98.96, "max": 99.83, "n": 3 }, "median_ttft_ms": { "median": 985.22, "min": 980.88, "max": 988.93, "n": 3 }, "p99_ttft_ms": { "median": 1053.19, "min": 1047.71, "max": 1088.49, "n": 3 }, "median_tpot_ms": { "median": 65.5, "min": 65.5, "max": 65.52, "n": 3 }, "p99_tpot_ms": { "median": 79.14, "min": 78.75, "max": 79.89, "n": 3 }, "p99_itl_ms": { "median": 70.27, "min": 69.95, "max": 70.74, "n": 3 }, "median_itl_ms": { "median": 65.39, "min": 65.37, "max": 65.4, "n": 3 }, "goodput_rps": { "median": 0.99, "min": 0.8, "max": 1.44, "n": 3 }, "request_throughput": { "median": 1.55, "min": 1.55, "max": 1.56, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1455.0, "max": 1500.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 166.1, 175.1, 180.3, 182.5, 184.5, 186.0, 186.8, 415.5, 420.6, 421.9, 424.1, 425.6, 427.5, 429.1, 430.5, 435.5, 438.8, 544.7, 545.4, 547.2, 974.0, 976.4, 978.3, 980.0, 984.6, 987.4, 988.0, 988.8, 989.1, 993.4, 994.6, 995.1, 995.8, 996.4, 997.1, 1000.3, 1022.0, 1031.7, 1032.8, 1033.2, 1037.3, 1039.0, 1040.9, 1047.2, 1052.4, 1053.5, 1077.6, 1079.3, 1082.2, 1086.9, 1088.7 ], "tpot_quantiles_ms": [ 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 73.2, 73.3, 73.4, 74.2, 74.3, 74.3, 74.4, 74.5, 74.5, 74.6, 75.4, 75.7, 75.9, 78.1, 78.3, 78.4, 78.7, 79.0, 79.3, 79.9 ] }, { "x": 1024.0, "output_throughput": { "median": 70.15, "min": 69.91, "max": 70.61, "n": 3 }, "median_ttft_ms": { "median": 2201.78, "min": 1943.14, "max": 2202.36, "n": 3 }, "p99_ttft_ms": { "median": 2977.57, "min": 2935.39, "max": 2997.9, "n": 3 }, "median_tpot_ms": { "median": 80.85, "min": 80.59, "max": 84.31, "n": 3 }, "p99_tpot_ms": { "median": 109.44, "min": 109.38, "max": 109.79, "n": 3 }, "p99_itl_ms": { "median": 200.62, "min": 81.13, "max": 210.73, "n": 3 }, "median_itl_ms": { "median": 69.07, "min": 68.99, "max": 69.08, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.1, "min": 1.09, "max": 1.1, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1425.0, "min": 1410.0, "max": 1425.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 366.8, 382.5, 389.8, 435.0, 437.9, 443.5, 447.5, 757.3, 766.4, 1582.7, 1587.4, 1593.1, 1596.4, 1602.3, 1613.8, 1621.5, 1624.1, 1629.0, 1633.5, 1640.0, 1641.7, 1865.5, 1923.0, 1934.8, 1941.5, 1942.9, 1950.7, 1959.9, 2761.3, 2773.0, 2775.2, 2780.6, 2795.4, 2810.1, 2814.7, 2830.5, 2836.4, 2841.8, 2845.0, 2850.2, 2853.6, 2871.6, 2896.9, 2900.4, 2903.0, 2906.9, 2915.2, 2934.8, 2955.1, 2977.5, 3005.7 ], "tpot_quantiles_ms": [ 69.0, 69.0, 69.0, 69.1, 69.1, 69.1, 69.1, 69.1, 69.2, 69.2, 69.2, 71.0, 71.1, 71.2, 71.2, 71.2, 71.2, 71.2, 71.3, 71.3, 71.4, 71.4, 71.7, 71.8, 84.1, 84.3, 84.4, 84.4, 84.4, 84.6, 89.8, 89.9, 89.9, 90.1, 90.1, 90.3, 90.5, 90.5, 90.6, 90.6, 90.7, 90.8, 101.9, 103.1, 107.5, 108.2, 108.6, 109.0, 109.3, 109.5, 110.0 ] }, { "x": 2048.0, "output_throughput": { "median": 46.69, "min": 46.17, "max": 47.91, "n": 3 }, "median_ttft_ms": { "median": 4346.21, "min": 4312.02, "max": 4449.93, "n": 3 }, "p99_ttft_ms": { "median": 9308.48, "min": 8481.04, "max": 9388.97, "n": 3 }, "median_tpot_ms": { "median": 107.94, "min": 106.78, "max": 108.25, "n": 3 }, "p99_tpot_ms": { "median": 187.71, "min": 175.13, "max": 188.99, "n": 3 }, "p99_itl_ms": { "median": 1239.64, "min": 1229.63, "max": 1243.1, "n": 3 }, "median_itl_ms": { "median": 73.94, "min": 73.85, "max": 73.95, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.73, "min": 0.72, "max": 0.75, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1297.5, "min": 1260.0, "max": 1320.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.189 }, "ttft_quantiles_ms": [ 654.4, 753.7, 770.5, 772.9, 810.7, 1013.2, 1087.4, 1918.8, 1947.3, 1968.3, 2004.5, 2207.0, 2894.9, 3108.3, 3149.2, 3160.2, 3169.2, 3171.8, 3204.4, 3215.6, 3225.8, 3239.8, 3646.8, 4299.2, 4321.3, 4369.3, 4376.1, 4405.1, 4435.8, 4449.1, 4453.1, 4464.9, 4482.8, 4786.4, 5086.2, 5412.2, 5441.6, 5501.7, 5520.5, 5566.4, 5589.4, 5603.8, 5626.4, 5674.2, 5998.0, 6072.7, 6516.4, 7316.4, 7618.9, 8481.6, 9389.2 ], "tpot_quantiles_ms": [ 73.8, 73.8, 73.9, 73.9, 73.9, 73.9, 74.0, 74.0, 74.0, 74.0, 74.1, 74.1, 74.1, 91.7, 91.8, 92.1, 92.2, 92.3, 92.4, 92.5, 92.6, 94.2, 96.8, 99.7, 102.7, 105.7, 110.9, 111.1, 111.5, 111.7, 112.1, 112.1, 112.3, 112.7, 113.9, 119.6, 128.4, 130.1, 131.1, 131.5, 132.1, 133.6, 138.8, 148.5, 149.4, 150.2, 150.7, 152.7, 162.8, 177.8, 204.6 ] }, { "x": 4096.0, "output_throughput": { "median": 20.8, "min": 20.23, "max": 21.17, "n": 3 }, "median_ttft_ms": { "median": 14706.12, "min": 14628.7, "max": 15135.64, "n": 3 }, "p99_ttft_ms": { "median": 30158.53, "min": 29607.79, "max": 30972.58, "n": 3 }, "median_tpot_ms": { "median": 119.48, "min": 116.16, "max": 120.96, "n": 3 }, "p99_tpot_ms": { "median": 217.41, "min": 214.58, "max": 227.2, "n": 3 }, "p99_itl_ms": { "median": 1178.69, "min": 1147.55, "max": 1315.21, "n": 3 }, "median_itl_ms": { "median": 76.15, "min": 76.09, "max": 76.16, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.33, "min": 0.32, "max": 0.33, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 930.0, "min": 922.5, "max": 975.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.618 }, "ttft_quantiles_ms": [ 1397.8, 3862.5, 5114.9, 6306.2, 6359.4, 6406.4, 6507.1, 6826.0, 6880.3, 7027.1, 7158.8, 8440.9, 9375.0, 10968.3, 11440.2, 12064.0, 12706.9, 12861.6, 13484.8, 13705.8, 13942.7, 14111.0, 14260.2, 14340.4, 14556.5, 14813.7, 15348.2, 15403.0, 15552.1, 15725.7, 16591.3, 16668.3, 16841.1, 17028.4, 17862.6, 19702.0, 20783.0, 22163.3, 22777.1, 23954.3, 24414.8, 24656.6, 25029.6, 25685.3, 26953.9, 27145.7, 27725.9, 28216.2, 29493.7, 30198.1, 31398.3 ], "tpot_quantiles_ms": [ 73.5, 76.0, 76.1, 76.2, 76.2, 76.2, 76.3, 76.3, 76.4, 76.4, 76.5, 84.9, 85.9, 86.2, 86.8, 89.1, 91.3, 93.1, 93.4, 94.7, 97.2, 103.4, 105.8, 106.5, 107.4, 118.5, 123.1, 125.2, 125.6, 126.4, 127.2, 127.9, 130.0, 132.0, 137.3, 140.9, 145.4, 147.5, 148.4, 150.7, 160.1, 162.5, 167.4, 171.1, 177.3, 180.9, 182.6, 201.2, 212.8, 217.6, 229.6 ] } ] }, "sglang-h2h": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 113.5, "min": 113.47, "max": 113.54, "n": 3 }, "median_ttft_ms": { "median": 353.79, "min": 351.25, "max": 361.04, "n": 3 }, "p99_ttft_ms": { "median": 482.92, "min": 475.0, "max": 503.28, "n": 3 }, "median_tpot_ms": { "median": 66.1, "min": 66.07, "max": 66.14, "n": 3 }, "p99_tpot_ms": { "median": 70.16, "min": 70.1, "max": 70.16, "n": 3 }, "p99_itl_ms": { "median": 72.43, "min": 71.57, "max": 74.91, "n": 3 }, "median_itl_ms": { "median": 64.15, "min": 64.13, "max": 64.16, "n": 3 }, "goodput_rps": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.77, "max": 1.77, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1500.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 84.8, 107.2, 109.2, 109.8, 110.4, 110.8, 112.0, 289.2, 291.3, 312.7, 317.0, 318.0, 323.1, 343.5, 344.5, 346.0, 348.7, 349.6, 349.9, 350.5, 350.9, 351.4, 351.8, 352.0, 352.8, 353.5, 354.1, 354.4, 355.1, 355.7, 356.4, 360.7, 362.3, 364.0, 370.3, 371.5, 445.7, 459.6, 462.5, 466.0, 466.8, 470.1, 470.7, 471.5, 473.2, 474.3, 474.9, 475.4, 477.5, 480.5, 542.0 ], "tpot_quantiles_ms": [ 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.7, 66.7, 66.7, 66.8, 67.2, 67.2, 68.3, 69.6, 69.9, 70.0, 70.0, 70.1, 70.2 ] }, { "x": 256.0, "output_throughput": { "median": 110.9, "min": 110.79, "max": 111.15, "n": 3 }, "median_ttft_ms": { "median": 470.06, "min": 323.66, "max": 473.83, "n": 3 }, "p99_ttft_ms": { "median": 856.32, "min": 689.81, "max": 900.81, "n": 3 }, "median_tpot_ms": { "median": 67.15, "min": 64.93, "max": 67.23, "n": 3 }, "p99_tpot_ms": { "median": 73.16, "min": 72.64, "max": 73.3, "n": 3 }, "p99_itl_ms": { "median": 71.56, "min": 71.52, "max": 71.96, "n": 3 }, "median_itl_ms": { "median": 64.73, "min": 64.72, "max": 64.75, "n": 3 }, "goodput_rps": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "request_throughput": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1485.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 88.4, 103.7, 111.9, 112.8, 113.7, 114.4, 117.2, 256.6, 281.5, 288.6, 289.9, 290.6, 291.4, 293.5, 294.6, 295.4, 296.8, 309.5, 316.0, 319.0, 320.2, 321.4, 322.4, 323.5, 325.6, 468.6, 469.3, 469.7, 471.9, 473.3, 473.9, 474.5, 474.9, 475.6, 476.9, 477.7, 478.5, 480.0, 480.5, 480.6, 481.0, 481.2, 481.5, 481.9, 489.4, 634.9, 636.0, 854.7, 856.2, 899.5, 901.0 ], "tpot_quantiles_ms": [ 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 67.2, 67.2, 67.2, 67.3, 67.4, 67.4, 67.4, 67.4, 67.5, 67.7, 67.7, 67.7, 67.8, 67.8, 67.8, 67.8, 67.9, 67.9, 68.3, 70.5, 70.5, 70.6, 70.7, 70.7, 70.8, 77.6 ] }, { "x": 1024.0, "output_throughput": { "median": 105.65, "min": 103.15, "max": 105.78, "n": 3 }, "median_ttft_ms": { "median": 466.04, "min": 463.78, "max": 493.18, "n": 3 }, "p99_ttft_ms": { "median": 1887.13, "min": 1725.09, "max": 1968.5, "n": 3 }, "median_tpot_ms": { "median": 66.75, "min": 66.74, "max": 66.79, "n": 3 }, "p99_tpot_ms": { "median": 89.71, "min": 89.62, "max": 90.87, "n": 3 }, "p99_itl_ms": { "median": 72.95, "min": 72.68, "max": 73.31, "n": 3 }, "median_itl_ms": { "median": 66.41, "min": 66.17, "max": 66.64, "n": 3 }, "goodput_rps": { "median": 1.52, "min": 1.28, "max": 1.52, "n": 3 }, "request_throughput": { "median": 1.65, "min": 1.61, "max": 1.65, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1440.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 89.7, 116.9, 118.0, 120.2, 121.2, 124.1, 127.4, 215.7, 218.2, 219.8, 222.2, 257.4, 259.9, 263.6, 437.1, 460.4, 461.9, 462.6, 463.1, 463.5, 464.0, 464.4, 464.7, 464.9, 465.1, 465.4, 465.8, 466.2, 466.2, 466.5, 466.8, 467.0, 467.3, 467.8, 468.4, 469.3, 470.4, 491.3, 493.0, 494.2, 494.9, 495.7, 501.9, 503.8, 657.2, 1237.0, 1376.3, 1677.9, 1829.9, 1886.6, 1968.7 ], "tpot_quantiles_ms": [ 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 68.8, 69.0, 69.9, 70.0, 70.1, 70.4, 70.6, 70.7, 71.2, 71.7, 72.1, 72.2, 72.3, 72.8, 85.1, 87.6, 96.5 ] }, { "x": 2048.0, "output_throughput": { "median": 100.38, "min": 99.62, "max": 102.45, "n": 3 }, "median_ttft_ms": { "median": 440.78, "min": 440.16, "max": 484.93, "n": 3 }, "p99_ttft_ms": { "median": 2316.63, "min": 2030.26, "max": 2685.73, "n": 3 }, "median_tpot_ms": { "median": 70.48, "min": 68.62, "max": 70.58, "n": 3 }, "p99_tpot_ms": { "median": 97.89, "min": 95.48, "max": 106.5, "n": 3 }, "p99_itl_ms": { "median": 150.62, "min": 73.79, "max": 152.94, "n": 3 }, "median_itl_ms": { "median": 69.1, "min": 68.1, "max": 69.14, "n": 3 }, "goodput_rps": { "median": 1.4, "min": 1.36, "max": 1.48, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.56, "max": 1.6, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1455.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 99.8, 113.9, 123.5, 164.4, 166.0, 166.4, 168.3, 222.9, 432.5, 433.9, 434.7, 435.7, 436.4, 437.2, 437.7, 438.4, 439.3, 439.4, 439.7, 440.2, 440.8, 441.3, 441.9, 442.1, 443.5, 444.3, 446.5, 470.6, 472.4, 479.9, 482.0, 483.9, 485.0, 486.0, 486.8, 487.4, 488.0, 488.8, 490.4, 491.8, 493.1, 494.8, 510.6, 516.1, 520.1, 524.8, 1638.5, 1876.6, 2030.1, 2684.1, 2686.1 ], "tpot_quantiles_ms": [ 68.0, 68.0, 68.0, 68.0, 68.0, 68.0, 68.0, 68.1, 68.1, 68.1, 68.1, 69.1, 69.2, 69.2, 69.2, 69.2, 69.2, 69.2, 69.2, 69.3, 69.3, 69.4, 69.4, 69.4, 69.4, 70.4, 70.5, 70.5, 70.5, 70.5, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.6, 70.7, 72.8, 73.6, 73.9, 74.4, 74.8, 74.9, 75.1, 76.2, 94.2, 110.0 ] }, { "x": 4096.0, "output_throughput": { "median": 88.36, "min": 88.29, "max": 88.93, "n": 3 }, "median_ttft_ms": { "median": 473.52, "min": 446.49, "max": 478.8, "n": 3 }, "p99_ttft_ms": { "median": 4274.95, "min": 3813.85, "max": 4792.76, "n": 3 }, "median_tpot_ms": { "median": 75.59, "min": 74.94, "max": 77.02, "n": 3 }, "p99_tpot_ms": { "median": 123.67, "min": 123.52, "max": 148.0, "n": 3 }, "p99_itl_ms": { "median": 161.53, "min": 81.87, "max": 214.22, "n": 3 }, "median_itl_ms": { "median": 74.05, "min": 73.91, "max": 74.16, "n": 3 }, "goodput_rps": { "median": 1.21, "min": 1.04, "max": 1.22, "n": 3 }, "request_throughput": { "median": 1.38, "min": 1.38, "max": 1.39, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1440.0, "min": 1440.0, "max": 1440.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 116.8, 163.5, 178.9, 188.2, 204.7, 216.2, 220.2, 363.1, 369.1, 375.3, 378.3, 404.7, 405.9, 408.2, 409.0, 411.2, 413.7, 415.1, 452.0, 454.3, 457.3, 471.0, 472.8, 474.3, 475.6, 477.3, 478.4, 479.5, 480.7, 483.7, 564.5, 578.6, 581.3, 582.5, 583.4, 585.0, 586.0, 586.7, 587.8, 588.6, 591.2, 592.4, 597.5, 612.8, 2090.9, 2093.4, 2564.4, 2672.5, 3814.2, 3943.9, 4837.6 ], "tpot_quantiles_ms": [ 71.8, 71.9, 72.0, 72.1, 73.2, 73.2, 73.2, 74.1, 74.1, 74.2, 74.2, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.3, 74.8, 74.8, 74.9, 75.0, 75.6, 75.6, 75.7, 76.3, 76.3, 76.4, 76.4, 76.5, 77.0, 77.1, 77.1, 77.2, 77.5, 77.7, 77.7, 77.8, 78.0, 78.2, 80.1, 80.3, 80.4, 80.6, 88.5, 98.8, 108.7, 119.5, 148.5 ] } ] }, "vllm": { "kv_pool_tokens": 24096, "points": [ { "x": 0.0, "output_throughput": { "median": 113.84, "min": 113.65, "max": 114.33, "n": 3 }, "median_ttft_ms": { "median": 407.88, "min": 370.74, "max": 444.84, "n": 3 }, "p99_ttft_ms": { "median": 484.79, "min": 480.58, "max": 490.12, "n": 3 }, "median_tpot_ms": { "median": 65.17, "min": 64.07, "max": 65.74, "n": 3 }, "p99_tpot_ms": { "median": 68.38, "min": 67.93, "max": 68.41, "n": 3 }, "p99_itl_ms": { "median": 133.61, "min": 69.14, "max": 167.24, "n": 3 }, "median_itl_ms": { "median": 64.04, "min": 64.04, "max": 64.07, "n": 3 }, "goodput_rps": { "median": 1.78, "min": 1.78, "max": 1.79, "n": 3 }, "request_throughput": { "median": 1.78, "min": 1.78, "max": 1.79, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1485.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 85.8, 150.7, 191.6, 213.0, 215.2, 216.0, 216.8, 252.8, 300.2, 341.4, 344.7, 364.9, 365.6, 366.8, 367.7, 370.2, 371.4, 371.8, 373.2, 375.3, 378.0, 390.9, 402.5, 403.5, 404.9, 407.7, 408.8, 409.1, 409.5, 410.7, 412.1, 442.2, 443.7, 444.8, 445.5, 446.4, 447.7, 450.8, 454.2, 455.0, 459.9, 462.1, 471.4, 474.2, 477.6, 478.1, 481.7, 482.5, 484.6, 486.5, 510.0 ], "tpot_quantiles_ms": [ 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 65.1, 65.1, 65.2, 65.2, 65.2, 65.2, 65.2, 65.3, 65.3, 65.7, 65.8, 65.9, 65.9, 65.9, 65.9, 65.9, 66.3, 67.1, 67.5, 67.7, 67.9, 68.2, 68.3, 68.4, 69.4 ] }, { "x": 256.0, "output_throughput": { "median": 111.46, "min": 111.44, "max": 111.67, "n": 3 }, "median_ttft_ms": { "median": 488.04, "min": 485.39, "max": 490.6, "n": 3 }, "p99_ttft_ms": { "median": 615.87, "min": 568.09, "max": 630.14, "n": 3 }, "median_tpot_ms": { "median": 64.83, "min": 64.83, "max": 64.83, "n": 3 }, "p99_tpot_ms": { "median": 70.1, "min": 70.02, "max": 70.38, "n": 3 }, "p99_itl_ms": { "median": 127.42, "min": 65.94, "max": 128.79, "n": 3 }, "median_itl_ms": { "median": 64.8, "min": 64.78, "max": 64.8, "n": 3 }, "goodput_rps": { "median": 1.74, "min": 1.74, "max": 1.74, "n": 3 }, "request_throughput": { "median": 1.74, "min": 1.74, "max": 1.74, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1485.0, "max": 1500.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 85.7, 156.6, 216.6, 217.8, 218.5, 219.3, 221.0, 274.7, 287.1, 295.2, 299.4, 301.5, 304.3, 305.6, 308.0, 336.4, 339.5, 343.8, 345.0, 347.7, 483.4, 484.6, 485.7, 486.3, 487.5, 488.7, 489.4, 490.0, 490.5, 490.8, 491.6, 492.4, 493.1, 493.7, 494.2, 494.7, 495.1, 495.7, 496.5, 496.8, 497.8, 500.5, 501.8, 503.0, 504.3, 505.4, 566.9, 568.3, 615.0, 629.1, 630.4 ], "tpot_quantiles_ms": [ 64.6, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.9, 64.9, 65.7, 65.8, 65.8, 65.9, 66.9, 67.1, 67.2, 67.6, 67.7, 67.8, 67.8, 67.9, 68.0, 68.1, 68.2, 68.3, 68.4, 68.9, 69.1, 69.2, 69.3, 69.5, 70.2, 70.4, 72.1 ] }, { "x": 1024.0, "output_throughput": { "median": 105.99, "min": 105.73, "max": 106.84, "n": 3 }, "median_ttft_ms": { "median": 395.89, "min": 394.77, "max": 474.55, "n": 3 }, "p99_ttft_ms": { "median": 1302.59, "min": 1011.96, "max": 1326.54, "n": 3 }, "median_tpot_ms": { "median": 68.9, "min": 66.78, "max": 69.04, "n": 3 }, "p99_tpot_ms": { "median": 79.91, "min": 77.02, "max": 80.28, "n": 3 }, "p99_itl_ms": { "median": 189.33, "min": 72.72, "max": 189.71, "n": 3 }, "median_itl_ms": { "median": 66.92, "min": 66.89, "max": 67.05, "n": 3 }, "goodput_rps": { "median": 1.52, "min": 1.51, "max": 1.53, "n": 3 }, "request_throughput": { "median": 1.66, "min": 1.65, "max": 1.67, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1455.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.4, 223.3, 226.7, 230.5, 231.7, 232.8, 235.9, 258.0, 267.1, 308.2, 310.0, 311.1, 313.4, 314.5, 315.6, 317.7, 389.1, 389.8, 390.7, 394.3, 394.5, 395.3, 399.3, 427.8, 465.5, 468.5, 470.8, 472.2, 475.1, 476.6, 477.9, 478.9, 480.1, 481.1, 481.9, 482.7, 484.7, 487.6, 488.6, 489.3, 490.8, 491.8, 492.7, 493.5, 504.3, 791.5, 794.9, 1011.4, 1022.5, 1302.5, 1326.8 ], "tpot_quantiles_ms": [ 66.6, 66.7, 66.7, 66.7, 66.7, 66.7, 66.8, 66.8, 66.8, 67.0, 67.1, 67.1, 67.1, 67.1, 67.3, 67.4, 67.5, 67.5, 67.5, 67.6, 67.8, 67.8, 68.1, 68.4, 68.5, 68.7, 68.8, 68.9, 69.0, 69.1, 69.3, 69.7, 69.9, 70.0, 70.1, 70.2, 70.2, 70.3, 70.4, 70.5, 70.5, 71.0, 71.2, 71.3, 71.5, 71.7, 72.1, 72.9, 75.3, 79.1, 83.3 ] }, { "x": 2048.0, "output_throughput": { "median": 100.18, "min": 99.5, "max": 100.91, "n": 3 }, "median_ttft_ms": { "median": 409.52, "min": 372.74, "max": 495.82, "n": 3 }, "p99_ttft_ms": { "median": 1766.94, "min": 1714.65, "max": 2147.98, "n": 3 }, "median_tpot_ms": { "median": 72.28, "min": 70.59, "max": 72.42, "n": 3 }, "p99_tpot_ms": { "median": 89.41, "min": 81.79, "max": 91.78, "n": 3 }, "p99_itl_ms": { "median": 150.77, "min": 128.25, "max": 165.36, "n": 3 }, "median_itl_ms": { "median": 70.06, "min": 69.68, "max": 70.29, "n": 3 }, "goodput_rps": { "median": 1.41, "min": 1.39, "max": 1.45, "n": 3 }, "request_throughput": { "median": 1.57, "min": 1.55, "max": 1.58, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1425.0, "max": 1485.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 104.9, 238.8, 248.2, 250.0, 251.6, 253.2, 254.5, 281.7, 286.8, 333.4, 335.5, 336.4, 360.3, 364.9, 366.5, 367.2, 372.2, 375.3, 398.7, 400.3, 404.4, 406.5, 407.2, 407.5, 408.6, 409.9, 412.6, 446.4, 451.8, 454.8, 460.2, 463.1, 468.7, 472.4, 485.8, 489.2, 495.0, 497.0, 500.8, 502.1, 504.0, 505.2, 507.5, 512.0, 513.9, 537.5, 1246.1, 1401.1, 1715.0, 1766.7, 2148.3 ], "tpot_quantiles_ms": [ 69.3, 69.3, 69.6, 69.7, 69.9, 70.1, 70.2, 70.2, 70.3, 70.4, 70.6, 70.7, 70.8, 70.9, 71.1, 71.2, 71.2, 71.3, 71.4, 71.5, 71.7, 71.8, 71.8, 72.1, 72.1, 72.2, 72.3, 72.4, 72.4, 72.4, 72.5, 72.5, 72.5, 72.8, 72.8, 73.0, 73.0, 73.2, 73.6, 73.6, 73.8, 74.0, 74.4, 74.5, 74.8, 74.9, 75.2, 75.9, 80.4, 88.8, 96.9 ] }, { "x": 4096.0, "output_throughput": { "median": 86.83, "min": 85.57, "max": 89.43, "n": 3 }, "median_ttft_ms": { "median": 374.73, "min": 362.7, "max": 510.66, "n": 3 }, "p99_ttft_ms": { "median": 4019.28, "min": 1949.66, "max": 4037.14, "n": 3 }, "median_tpot_ms": { "median": 80.76, "min": 77.36, "max": 81.8, "n": 3 }, "p99_tpot_ms": { "median": 115.47, "min": 99.55, "max": 125.63, "n": 3 }, "p99_itl_ms": { "median": 229.66, "min": 140.16, "max": 231.75, "n": 3 }, "median_itl_ms": { "median": 76.8, "min": 76.13, "max": 78.0, "n": 3 }, "goodput_rps": { "median": 1.17, "min": 1.09, "max": 1.19, "n": 3 }, "request_throughput": { "median": 1.36, "min": 1.34, "max": 1.4, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1395.0, "max": 1410.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 116.4, 255.4, 269.1, 282.9, 286.5, 288.2, 290.5, 294.1, 301.6, 303.3, 312.9, 315.8, 328.7, 332.5, 333.8, 339.2, 341.0, 342.9, 344.6, 354.9, 363.9, 374.5, 380.5, 454.0, 458.8, 461.3, 467.0, 468.9, 469.7, 470.7, 471.7, 472.2, 473.8, 493.3, 494.2, 496.4, 503.6, 510.2, 512.7, 514.0, 515.5, 518.8, 522.7, 756.8, 1505.2, 1834.4, 1900.9, 2096.0, 2778.8, 4019.1, 4037.5 ], "tpot_quantiles_ms": [ 73.6, 75.1, 75.4, 76.3, 77.0, 77.1, 77.1, 77.1, 77.2, 77.3, 77.4, 77.6, 77.7, 77.8, 78.5, 78.6, 78.8, 78.9, 79.0, 79.1, 79.2, 79.6, 80.0, 80.0, 80.2, 80.5, 80.7, 80.9, 80.9, 81.0, 81.1, 81.2, 81.4, 81.5, 81.8, 81.8, 82.3, 82.5, 82.8, 82.8, 83.9, 84.5, 84.7, 85.0, 85.3, 85.7, 86.5, 88.6, 96.5, 106.1, 131.4 ] } ] }, "small-pool": { "kv_pool_tokens": 6559, "points": [ { "x": 0.0, "output_throughput": { "median": 113.46, "min": 113.4, "max": 114.16, "n": 3 }, "median_ttft_ms": { "median": 354.05, "min": 346.41, "max": 355.32, "n": 3 }, "p99_ttft_ms": { "median": 477.34, "min": 465.08, "max": 502.44, "n": 3 }, "median_tpot_ms": { "median": 66.13, "min": 65.95, "max": 66.14, "n": 3 }, "p99_tpot_ms": { "median": 70.17, "min": 69.82, "max": 70.22, "n": 3 }, "p99_itl_ms": { "median": 72.9, "min": 72.69, "max": 180.6, "n": 3 }, "median_itl_ms": { "median": 64.18, "min": 64.04, "max": 64.2, "n": 3 }, "goodput_rps": { "median": 1.77, "min": 1.77, "max": 1.78, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.77, "max": 1.78, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1500.0, "min": 1500.0, "max": 1560.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 83.6, 106.3, 108.0, 109.0, 109.8, 111.4, 111.9, 289.4, 331.9, 335.8, 342.5, 343.3, 343.7, 344.6, 345.6, 347.6, 348.6, 349.2, 349.6, 350.1, 350.4, 351.0, 351.9, 352.6, 353.3, 354.0, 354.4, 355.0, 355.3, 355.7, 356.0, 356.3, 356.9, 357.8, 358.5, 359.6, 361.5, 362.9, 387.1, 388.4, 453.3, 458.8, 459.7, 464.1, 467.4, 469.3, 472.1, 473.7, 476.6, 477.5, 543.5 ], "tpot_quantiles_ms": [ 64.0, 64.1, 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.3, 64.4, 65.9, 66.0, 66.0, 66.0, 66.0, 66.0, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.1, 66.2, 66.2, 66.2, 66.2, 66.2, 66.2, 66.3, 66.3, 66.3, 66.4, 68.2, 68.6, 69.7, 69.9, 70.1, 70.1, 70.3 ] }, { "x": 256.0, "output_throughput": { "median": 111.0, "min": 110.83, "max": 111.05, "n": 3 }, "median_ttft_ms": { "median": 470.24, "min": 324.03, "max": 475.58, "n": 3 }, "p99_ttft_ms": { "median": 790.14, "min": 696.0, "max": 905.09, "n": 3 }, "median_tpot_ms": { "median": 67.18, "min": 67.08, "max": 67.69, "n": 3 }, "p99_tpot_ms": { "median": 72.71, "min": 72.7, "max": 73.32, "n": 3 }, "p99_itl_ms": { "median": 71.57, "min": 71.45, "max": 78.79, "n": 3 }, "median_itl_ms": { "median": 64.77, "min": 64.76, "max": 64.79, "n": 3 }, "goodput_rps": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "request_throughput": { "median": 1.73, "min": 1.73, "max": 1.74, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 91.4, 103.0, 112.4, 113.0, 114.3, 115.8, 119.6, 256.1, 278.2, 287.9, 289.2, 289.7, 290.6, 292.2, 292.9, 294.2, 294.9, 295.4, 315.3, 317.0, 319.1, 321.0, 322.9, 324.1, 464.5, 465.5, 469.8, 470.3, 470.7, 471.5, 474.3, 475.0, 475.6, 476.0, 476.4, 477.0, 477.6, 478.1, 479.3, 479.9, 480.9, 481.5, 482.2, 483.0, 493.1, 564.9, 566.5, 644.7, 790.0, 903.7, 905.3 ], "tpot_quantiles_ms": [ 64.7, 64.7, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.8, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 64.9, 65.0, 65.0, 65.0, 65.0, 67.1, 67.1, 67.2, 67.3, 67.4, 67.4, 67.4, 67.4, 67.7, 67.7, 67.7, 67.8, 67.8, 67.8, 67.9, 67.9, 67.9, 68.0, 68.5, 68.5, 70.4, 70.5, 70.6, 70.7, 70.8, 70.8, 77.6 ] }, { "x": 1024.0, "output_throughput": { "median": 94.97, "min": 94.95, "max": 95.86, "n": 3 }, "median_ttft_ms": { "median": 469.5, "min": 469.28, "max": 472.19, "n": 3 }, "p99_ttft_ms": { "median": 5769.23, "min": 5731.62, "max": 5892.35, "n": 3 }, "median_tpot_ms": { "median": 66.79, "min": 66.75, "max": 66.81, "n": 3 }, "p99_tpot_ms": { "median": 78.48, "min": 78.45, "max": 79.6, "n": 3 }, "p99_itl_ms": { "median": 74.33, "min": 73.17, "max": 77.85, "n": 3 }, "median_itl_ms": { "median": 66.58, "min": 66.57, "max": 66.61, "n": 3 }, "goodput_rps": { "median": 1.37, "min": 1.27, "max": 1.38, "n": 3 }, "request_throughput": { "median": 1.48, "min": 1.48, "max": 1.5, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1470.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 90.3, 116.6, 118.1, 121.5, 217.2, 231.5, 237.3, 239.0, 240.0, 241.6, 244.2, 245.7, 257.9, 259.6, 263.9, 387.8, 395.0, 440.4, 461.5, 464.5, 466.5, 467.0, 468.0, 468.8, 469.3, 469.7, 470.5, 471.7, 473.8, 474.6, 475.5, 477.4, 479.6, 481.6, 485.0, 487.5, 493.8, 495.4, 510.9, 653.0, 679.2, 682.8, 717.2, 729.8, 731.1, 733.0, 1228.6, 1319.7, 5731.1, 5769.1, 5892.6 ], "tpot_quantiles_ms": [ 64.6, 65.2, 65.8, 66.1, 66.1, 66.2, 66.2, 66.2, 66.3, 66.3, 66.6, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.7, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.8, 66.9, 68.2, 69.3, 69.4, 69.8, 70.2, 70.4, 70.4, 70.4, 70.4, 70.5, 70.5, 70.5, 70.6, 71.6, 71.6, 72.4, 72.8, 74.2, 76.3, 87.4 ] }, { "x": 2048.0, "output_throughput": { "median": 25.87, "min": 24.69, "max": 29.49, "n": 3 }, "median_ttft_ms": { "median": 12576.16, "min": 11622.68, "max": 15608.52, "n": 3 }, "p99_ttft_ms": { "median": 24833.24, "min": 22504.41, "max": 24887.65, "n": 3 }, "median_tpot_ms": { "median": 66.11, "min": 65.91, "max": 66.26, "n": 3 }, "p99_tpot_ms": { "median": 89.04, "min": 77.31, "max": 92.96, "n": 3 }, "p99_itl_ms": { "median": 104.27, "min": 68.92, "max": 104.28, "n": 3 }, "median_itl_ms": { "median": 65.89, "min": 65.8, "max": 65.9, "n": 3 }, "goodput_rps": { "median": 0.01, "min": 0.01, "max": 0.02, "n": 3 }, "request_throughput": { "median": 0.4, "min": 0.39, "max": 0.46, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1155.0, "max": 1380.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.508 }, "ttft_quantiles_ms": [ 100.1, 456.1, 1107.0, 5458.3, 5710.2, 5927.4, 6806.8, 8129.2, 9768.6, 10011.1, 10282.2, 10413.2, 10495.6, 10818.5, 10917.7, 11058.4, 11136.0, 11257.3, 11339.1, 11413.9, 11537.0, 11822.9, 12069.6, 12419.9, 12502.5, 12615.8, 13982.6, 14885.2, 15219.7, 15495.3, 15537.4, 15836.1, 15992.4, 16465.2, 16544.4, 16693.1, 16809.6, 16870.7, 17065.0, 17227.1, 17442.0, 17602.9, 17988.1, 18576.5, 19252.9, 19871.6, 20320.1, 22044.7, 23962.1, 24574.2, 25099.9 ], "tpot_quantiles_ms": [ 64.3, 64.4, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.5, 65.6, 65.6, 65.6, 65.6, 65.6, 65.7, 65.8, 65.8, 65.8, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 66.0, 66.0, 66.0, 66.1, 66.1, 66.3, 66.3, 66.3, 66.4, 66.6, 66.6, 66.7, 67.1, 67.5, 68.3, 68.4, 69.3, 69.4, 69.5, 70.4, 71.9, 72.4, 74.2, 77.3, 78.0, 86.4, 104.2 ] }, { "x": 4096.0, "output_throughput": { "median": 10.87, "min": 10.82, "max": 11.08, "n": 3 }, "median_ttft_ms": { "median": 40126.15, "min": 35819.4, "max": 43671.4, "n": 3 }, "p99_ttft_ms": { "median": 67558.17, "min": 59110.4, "max": 68639.85, "n": 3 }, "median_tpot_ms": { "median": 66.33, "min": 65.73, "max": 66.35, "n": 3 }, "p99_tpot_ms": { "median": 93.33, "min": 83.02, "max": 97.29, "n": 3 }, "p99_itl_ms": { "median": 97.31, "min": 83.9, "max": 99.03, "n": 3 }, "median_itl_ms": { "median": 65.97, "min": 65.71, "max": 66.24, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.01, "n": 3 }, "request_throughput": { "median": 0.17, "min": 0.17, "max": 0.17, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 390.0, "min": 360.0, "max": 405.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.79 }, "ttft_quantiles_ms": [ 121.5, 214.3, 5528.7, 11026.4, 21651.9, 26980.5, 27026.0, 28459.1, 29455.0, 29883.3, 30193.5, 32310.7, 33126.2, 33664.0, 34224.9, 34776.3, 35099.9, 35282.7, 36223.7, 36373.2, 37274.5, 38058.5, 38714.2, 38807.3, 38958.0, 39495.0, 39995.8, 40534.6, 41039.6, 41617.2, 43532.4, 44852.7, 45207.2, 46432.7, 46617.0, 46941.2, 47820.2, 48677.4, 49162.9, 49910.8, 50061.8, 50227.8, 51187.7, 51732.0, 55055.2, 57093.9, 58426.7, 60591.5, 63470.0, 67148.7, 69543.9 ], "tpot_quantiles_ms": [ 61.5, 62.4, 62.4, 62.4, 62.4, 62.4, 62.5, 62.5, 62.5, 62.5, 62.8, 63.1, 63.1, 63.6, 64.3, 64.7, 65.2, 65.5, 65.5, 65.6, 65.6, 65.6, 65.7, 66.0, 66.2, 66.3, 66.3, 66.3, 66.4, 66.4, 66.5, 66.5, 66.6, 66.7, 66.8, 66.9, 67.1, 67.2, 67.5, 67.7, 71.3, 74.7, 75.1, 75.8, 81.8, 82.6, 83.1, 85.5, 87.4, 88.7, 97.3 ] } ] }, "fp8-small-pool": { "kv_pool_tokens": 107299, "points": [ { "x": 0.0, "output_throughput": { "median": 180.12, "min": 180.01, "max": 180.29, "n": 3 }, "median_ttft_ms": { "median": 255.34, "min": 254.37, "max": 255.59, "n": 3 }, "p99_ttft_ms": { "median": 330.98, "min": 328.19, "max": 340.82, "n": 3 }, "median_tpot_ms": { "median": 41.07, "min": 41.06, "max": 41.11, "n": 3 }, "p99_tpot_ms": { "median": 44.01, "min": 43.95, "max": 44.04, "n": 3 }, "p99_itl_ms": { "median": 83.65, "min": 50.02, "max": 113.32, "n": 3 }, "median_itl_ms": { "median": 39.83, "min": 39.83, "max": 39.85, "n": 3 }, "goodput_rps": { "median": 2.81, "min": 2.81, "max": 2.82, "n": 3 }, "request_throughput": { "median": 2.81, "min": 2.81, "max": 2.82, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1260.0, "min": 1230.0, "max": 1290.0, "n": 3 }, "max_temp_c": 80.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 72.5, 76.6, 77.2, 77.6, 77.9, 80.4, 86.6, 221.0, 222.5, 230.8, 247.3, 248.1, 248.4, 249.0, 249.4, 249.9, 251.0, 251.6, 252.2, 252.5, 252.9, 253.2, 253.8, 254.0, 254.4, 254.7, 255.3, 255.9, 256.4, 256.7, 258.4, 269.8, 273.9, 276.3, 278.5, 281.1, 283.8, 286.1, 287.4, 315.2, 318.7, 319.7, 320.4, 321.6, 322.4, 322.8, 325.1, 327.0, 327.4, 329.0, 363.7 ], "tpot_quantiles_ms": [ 39.8, 39.8, 39.8, 39.8, 39.8, 39.8, 39.9, 39.9, 39.9, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.0, 40.1, 40.1, 40.6, 40.6, 40.7, 41.0, 41.0, 41.1, 41.1, 41.1, 41.1, 41.1, 41.1, 41.1, 41.1, 41.1, 41.1, 41.2, 41.2, 41.2, 41.2, 41.2, 41.2, 41.2, 41.2, 41.3, 41.5, 41.5, 42.9, 43.8, 43.8, 43.9, 43.9, 44.0, 44.1 ] }, { "x": 256.0, "output_throughput": { "median": 175.23, "min": 174.89, "max": 175.52, "n": 3 }, "median_ttft_ms": { "median": 323.34, "min": 256.17, "max": 328.04, "n": 3 }, "p99_ttft_ms": { "median": 698.28, "min": 642.85, "max": 757.44, "n": 3 }, "median_tpot_ms": { "median": 40.35, "min": 40.35, "max": 41.53, "n": 3 }, "p99_tpot_ms": { "median": 46.55, "min": 46.16, "max": 46.84, "n": 3 }, "p99_itl_ms": { "median": 46.99, "min": 46.84, "max": 48.14, "n": 3 }, "median_itl_ms": { "median": 40.18, "min": 40.17, "max": 40.2, "n": 3 }, "goodput_rps": { "median": 2.74, "min": 2.73, "max": 2.74, "n": 3 }, "request_throughput": { "median": 2.74, "min": 2.73, "max": 2.74, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1260.0, "min": 1245.0, "max": 1290.0, "n": 3 }, "max_temp_c": 80.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 75.8, 79.5, 81.4, 82.1, 82.4, 82.7, 84.4, 192.4, 196.7, 197.7, 198.5, 199.1, 199.8, 201.6, 202.8, 203.9, 205.7, 208.2, 224.8, 227.1, 228.7, 232.4, 249.1, 255.4, 257.6, 260.1, 318.8, 322.4, 327.0, 327.6, 327.9, 328.4, 328.7, 329.1, 330.3, 330.7, 331.9, 332.4, 332.9, 333.4, 334.9, 335.6, 336.6, 338.7, 343.7, 640.9, 642.4, 696.7, 698.1, 756.1, 757.7 ], "tpot_quantiles_ms": [ 40.2, 40.2, 40.2, 40.2, 40.2, 40.2, 40.2, 40.2, 40.2, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.3, 40.4, 40.4, 40.4, 40.4, 41.4, 41.5, 41.5, 41.6, 41.9, 41.9, 42.0, 42.1, 42.2, 42.3, 42.3, 42.4, 42.4, 42.4, 42.4, 42.4, 42.5, 42.5, 44.0, 44.2, 44.3, 44.3, 44.4, 44.5, 50.9 ] }, { "x": 1024.0, "output_throughput": { "median": 164.11, "min": 159.11, "max": 164.95, "n": 3 }, "median_ttft_ms": { "median": 348.54, "min": 346.15, "max": 351.72, "n": 3 }, "p99_ttft_ms": { "median": 1565.27, "min": 1457.12, "max": 1592.64, "n": 3 }, "median_tpot_ms": { "median": 41.44, "min": 41.38, "max": 41.48, "n": 3 }, "p99_tpot_ms": { "median": 55.44, "min": 54.75, "max": 60.79, "n": 3 }, "p99_itl_ms": { "median": 47.93, "min": 47.59, "max": 48.13, "n": 3 }, "median_itl_ms": { "median": 41.28, "min": 41.26, "max": 41.33, "n": 3 }, "goodput_rps": { "median": 2.28, "min": 2.02, "max": 2.46, "n": 3 }, "request_throughput": { "median": 2.56, "min": 2.49, "max": 2.58, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1260.0, "min": 1215.0, "max": 1290.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 69.7, 84.3, 87.0, 89.1, 89.5, 90.4, 102.8, 149.8, 152.4, 155.4, 165.9, 180.4, 181.7, 332.5, 333.8, 334.7, 340.6, 341.3, 342.5, 345.0, 346.7, 347.2, 347.7, 347.8, 348.4, 348.7, 349.0, 349.3, 349.8, 350.3, 350.8, 351.4, 352.0, 352.5, 352.7, 353.5, 354.7, 355.0, 355.1, 356.2, 357.5, 358.7, 359.7, 442.0, 925.1, 1033.4, 1065.5, 1159.7, 1457.0, 1565.1, 1592.9 ], "tpot_quantiles_ms": [ 41.2, 41.2, 41.3, 41.3, 41.3, 41.3, 41.3, 41.3, 41.3, 41.3, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.4, 41.5, 41.5, 41.5, 41.5, 41.5, 41.5, 41.5, 42.0, 42.0, 44.0, 44.2, 44.4, 44.5, 44.5, 44.6, 45.4, 45.6, 45.6, 45.6, 45.7, 49.7, 49.7, 49.8, 52.7, 65.2 ] }, { "x": 2048.0, "output_throughput": { "median": 155.07, "min": 153.64, "max": 158.0, "n": 3 }, "median_ttft_ms": { "median": 380.11, "min": 365.42, "max": 380.24, "n": 3 }, "p99_ttft_ms": { "median": 1916.66, "min": 1652.76, "max": 2179.55, "n": 3 }, "median_tpot_ms": { "median": 43.33, "min": 43.02, "max": 44.73, "n": 3 }, "p99_tpot_ms": { "median": 62.04, "min": 57.72, "max": 67.31, "n": 3 }, "p99_itl_ms": { "median": 49.27, "min": 47.23, "max": 49.73, "n": 3 }, "median_itl_ms": { "median": 42.96, "min": 42.9, "max": 42.97, "n": 3 }, "goodput_rps": { "median": 2.2, "min": 2.16, "max": 2.21, "n": 3 }, "request_throughput": { "median": 2.42, "min": 2.4, "max": 2.47, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1245.0, "min": 1245.0, "max": 1260.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 89.3, 90.7, 92.9, 115.2, 129.4, 129.9, 144.4, 156.9, 159.5, 238.7, 244.3, 246.1, 246.9, 247.7, 248.0, 249.2, 251.2, 291.8, 293.3, 354.3, 356.5, 363.3, 364.9, 365.1, 365.7, 366.0, 366.7, 367.4, 368.3, 369.8, 376.9, 379.1, 380.6, 385.9, 386.6, 388.1, 389.3, 390.1, 390.9, 391.6, 392.3, 393.0, 393.7, 395.6, 401.7, 959.0, 1097.2, 1651.6, 1915.9, 1963.6, 2179.8 ], "tpot_quantiles_ms": [ 42.8, 42.9, 42.9, 42.9, 42.9, 42.9, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.0, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 43.1, 44.6, 44.7, 45.2, 45.3, 45.3, 45.3, 45.3, 45.4, 45.4, 45.4, 45.9, 46.3, 46.4, 46.8, 47.1, 47.2, 47.2, 47.3, 47.5, 51.8, 62.3, 75.8 ] }, { "x": 4096.0, "output_throughput": { "median": 134.81, "min": 133.54, "max": 136.45, "n": 3 }, "median_ttft_ms": { "median": 371.7, "min": 370.22, "max": 424.2, "n": 3 }, "p99_ttft_ms": { "median": 3482.57, "min": 2794.52, "max": 4087.11, "n": 3 }, "median_tpot_ms": { "median": 47.65, "min": 46.37, "max": 47.74, "n": 3 }, "p99_tpot_ms": { "median": 88.21, "min": 77.61, "max": 88.82, "n": 3 }, "p99_itl_ms": { "median": 101.51, "min": 50.28, "max": 124.14, "n": 3 }, "median_itl_ms": { "median": 46.07, "min": 46.06, "max": 46.16, "n": 3 }, "goodput_rps": { "median": 1.83, "min": 1.65, "max": 1.9, "n": 3 }, "request_throughput": { "median": 2.11, "min": 2.09, "max": 2.13, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1200.0, "min": 1200.0, "max": 1200.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 108.3, 141.2, 149.8, 170.2, 171.3, 173.5, 180.0, 223.7, 225.0, 225.8, 226.5, 227.3, 228.5, 255.3, 265.6, 267.0, 268.4, 269.7, 312.5, 358.8, 366.4, 367.7, 369.5, 370.3, 372.0, 373.1, 395.8, 401.7, 405.1, 423.1, 423.8, 424.6, 425.2, 425.7, 426.2, 427.1, 428.4, 450.2, 452.9, 454.6, 455.9, 459.4, 465.9, 1154.6, 1528.6, 1729.6, 1733.4, 2177.8, 2951.0, 3481.7, 4087.5 ], "tpot_quantiles_ms": [ 45.8, 46.0, 46.1, 46.1, 46.1, 46.1, 46.1, 46.2, 46.2, 46.2, 46.2, 46.2, 46.3, 46.3, 46.3, 46.4, 46.4, 46.8, 47.0, 47.1, 47.2, 47.5, 47.5, 47.6, 47.6, 47.7, 47.7, 47.8, 47.8, 48.6, 48.7, 48.8, 48.8, 48.9, 49.6, 49.8, 49.8, 50.0, 50.0, 50.1, 50.2, 50.5, 50.6, 50.8, 50.8, 50.9, 54.6, 65.9, 71.2, 76.4, 108.3 ] } ] }, "vllm-no-prefix-cache": { "kv_pool_tokens": 24096, "points": [ { "x": 0.0, "output_throughput": { "median": 113.57, "min": 113.48, "max": 114.25, "n": 3 }, "median_ttft_ms": { "median": 392.13, "min": 373.68, "max": 407.57, "n": 3 }, "p99_ttft_ms": { "median": 492.09, "min": 469.75, "max": 510.21, "n": 3 }, "median_tpot_ms": { "median": 65.26, "min": 64.99, "max": 65.8, "n": 3 }, "p99_tpot_ms": { "median": 68.39, "min": 68.06, "max": 68.43, "n": 3 }, "p99_itl_ms": { "median": 134.68, "min": 129.24, "max": 172.94, "n": 3 }, "median_itl_ms": { "median": 64.06, "min": 63.93, "max": 64.07, "n": 3 }, "goodput_rps": { "median": 1.77, "min": 1.77, "max": 1.79, "n": 3 }, "request_throughput": { "median": 1.77, "min": 1.77, "max": 1.79, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1470.0, "max": 1575.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 99.3, 149.8, 209.8, 210.6, 213.2, 214.8, 216.4, 306.8, 325.7, 344.0, 361.3, 365.4, 366.9, 370.1, 371.7, 373.6, 383.2, 384.1, 385.2, 388.2, 390.1, 391.5, 391.9, 392.3, 392.7, 393.1, 393.8, 396.4, 397.4, 399.2, 402.2, 406.3, 407.3, 408.3, 409.2, 410.4, 411.9, 414.1, 456.2, 460.9, 467.6, 474.5, 478.9, 480.3, 481.3, 481.8, 483.6, 484.2, 486.5, 487.2, 549.7 ], "tpot_quantiles_ms": [ 63.9, 63.9, 63.9, 63.9, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.0, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 64.1, 65.0, 65.0, 65.0, 65.0, 65.0, 65.0, 65.1, 65.1, 65.1, 65.2, 65.2, 65.3, 65.3, 65.3, 65.3, 65.3, 65.4, 65.4, 65.8, 65.8, 65.8, 65.9, 65.9, 65.9, 65.9, 66.1, 67.6, 67.8, 68.3, 68.3, 68.4, 68.5, 70.9 ] }, { "x": 256.0, "output_throughput": { "median": 99.1, "min": 98.97, "max": 99.56, "n": 3 }, "median_ttft_ms": { "median": 752.77, "min": 746.5, "max": 827.83, "n": 3 }, "p99_ttft_ms": { "median": 1096.04, "min": 1082.31, "max": 1103.02, "n": 3 }, "median_tpot_ms": { "median": 70.17, "min": 68.29, "max": 70.25, "n": 3 }, "p99_tpot_ms": { "median": 77.63, "min": 77.29, "max": 77.65, "n": 3 }, "p99_itl_ms": { "median": 322.89, "min": 255.21, "max": 378.79, "n": 3 }, "median_itl_ms": { "median": 65.27, "min": 65.27, "max": 65.29, "n": 3 }, "goodput_rps": { "median": 0.97, "min": 0.97, "max": 1.12, "n": 3 }, "request_throughput": { "median": 1.55, "min": 1.55, "max": 1.56, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1462.5, "min": 1455.0, "max": 1470.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 176.9, 292.7, 297.3, 298.9, 299.9, 301.8, 305.0, 731.8, 734.0, 738.0, 739.0, 740.1, 740.8, 741.4, 745.0, 746.8, 747.3, 747.7, 748.1, 749.1, 750.1, 751.8, 755.7, 758.7, 768.2, 770.0, 813.3, 824.5, 825.9, 827.5, 828.7, 833.8, 835.5, 1000.8, 1005.1, 1007.8, 1019.0, 1023.4, 1033.6, 1044.3, 1053.8, 1056.9, 1057.7, 1066.1, 1067.9, 1068.7, 1071.7, 1073.1, 1082.1, 1095.9, 1103.4 ], "tpot_quantiles_ms": [ 65.2, 65.2, 65.2, 65.2, 65.2, 65.2, 65.2, 65.2, 65.2, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 66.2, 66.4, 67.4, 68.2, 68.3, 68.3, 68.3, 68.3, 68.3, 68.4, 68.4, 69.4, 69.5, 69.5, 70.2, 70.2, 70.3, 70.3, 70.3, 70.3, 70.3, 70.4, 70.4, 70.5, 70.6, 71.3, 71.4, 71.5, 71.6, 76.3, 76.4, 77.1, 77.5, 77.7, 78.5, 79.2 ] }, { "x": 1024.0, "output_throughput": { "median": 70.88, "min": 70.86, "max": 70.95, "n": 3 }, "median_ttft_ms": { "median": 1213.27, "min": 1212.38, "max": 1214.18, "n": 3 }, "p99_ttft_ms": { "median": 2852.85, "min": 2827.56, "max": 2864.82, "n": 3 }, "median_tpot_ms": { "median": 94.93, "min": 94.77, "max": 95.0, "n": 3 }, "p99_tpot_ms": { "median": 106.34, "min": 106.32, "max": 106.48, "n": 3 }, "p99_itl_ms": { "median": 601.7, "min": 598.27, "max": 603.84, "n": 3 }, "median_itl_ms": { "median": 69.02, "min": 69.02, "max": 69.03, "n": 3 }, "goodput_rps": { "median": 0.05, "min": 0.03, "max": 0.05, "n": 3 }, "request_throughput": { "median": 1.11, "min": 1.11, "max": 1.11, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1395.0, "max": 1410.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 385.0, 529.4, 532.1, 537.9, 542.0, 546.7, 554.6, 852.9, 859.2, 869.2, 881.4, 886.8, 965.0, 1173.0, 1180.3, 1187.5, 1190.5, 1194.3, 1197.7, 1203.2, 1204.9, 1207.4, 1210.1, 1210.6, 1212.3, 1213.4, 1216.0, 1219.5, 1222.0, 1228.0, 1372.9, 1386.3, 1395.5, 1413.9, 1504.5, 1515.7, 1557.6, 1582.0, 1584.6, 1591.3, 1604.6, 1747.2, 1898.0, 1918.6, 1975.4, 1984.9, 2174.3, 2194.8, 2206.0, 2795.7, 2982.5 ], "tpot_quantiles_ms": [ 68.7, 74.2, 79.7, 81.0, 84.4, 84.4, 84.7, 85.2, 85.8, 87.3, 87.7, 87.8, 88.0, 89.1, 89.6, 89.8, 89.9, 90.1, 90.3, 90.5, 91.3, 91.7, 92.2, 93.9, 94.7, 95.0, 95.1, 95.2, 95.3, 95.5, 95.7, 95.7, 95.8, 95.9, 95.9, 96.1, 96.9, 97.3, 98.0, 99.4, 100.0, 100.6, 101.3, 102.8, 104.1, 104.4, 105.0, 105.6, 106.0, 107.5, 110.3 ] }, { "x": 2048.0, "output_throughput": { "median": 46.58, "min": 46.17, "max": 46.7, "n": 3 }, "median_ttft_ms": { "median": 2549.7, "min": 2545.69, "max": 2562.85, "n": 3 }, "p99_ttft_ms": { "median": 5214.79, "min": 5214.5, "max": 5393.28, "n": 3 }, "median_tpot_ms": { "median": 119.75, "min": 119.36, "max": 126.88, "n": 3 }, "p99_tpot_ms": { "median": 185.46, "min": 180.25, "max": 190.89, "n": 3 }, "p99_itl_ms": { "median": 1026.05, "min": 961.18, "max": 1049.22, "n": 3 }, "median_itl_ms": { "median": 74.15, "min": 74.1, "max": 74.15, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.73, "min": 0.72, "max": 0.73, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1035.0, "min": 1027.5, "max": 1035.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.252 }, "ttft_quantiles_ms": [ 1229.7, 1387.2, 1406.5, 1420.1, 1432.4, 1831.6, 1845.5, 1961.3, 1982.0, 1993.0, 2009.3, 2286.2, 2450.1, 2480.3, 2483.7, 2496.5, 2511.2, 2513.6, 2518.0, 2521.8, 2526.0, 2531.6, 2539.4, 2543.0, 2547.3, 2550.2, 2552.2, 2558.2, 2562.6, 2573.5, 2576.5, 2582.9, 2589.5, 2614.4, 2665.7, 2761.8, 2957.3, 3193.7, 3511.2, 3531.3, 3633.9, 3710.9, 3737.1, 3780.7, 4349.0, 4452.1, 4628.2, 4774.5, 4979.4, 5169.8, 5617.8 ], "tpot_quantiles_ms": [ 73.3, 88.0, 106.8, 117.5, 117.7, 118.0, 118.0, 118.1, 118.3, 118.5, 118.6, 118.6, 118.7, 118.7, 118.8, 119.0, 119.3, 119.3, 119.4, 119.5, 119.8, 119.9, 120.3, 120.4, 120.6, 121.3, 125.8, 126.8, 126.9, 127.2, 127.9, 129.2, 130.6, 132.6, 134.5, 135.4, 136.1, 136.9, 137.8, 138.1, 140.1, 143.6, 145.5, 149.6, 153.9, 156.9, 160.9, 167.1, 173.7, 182.0, 199.4 ] }, { "x": 4096.0, "output_throughput": { "median": 19.57, "min": 19.5, "max": 19.7, "n": 3 }, "median_ttft_ms": { "median": 13353.32, "min": 13252.44, "max": 13466.36, "n": 3 }, "p99_ttft_ms": { "median": 21138.92, "min": 21076.34, "max": 21499.96, "n": 3 }, "median_tpot_ms": { "median": 192.93, "min": 191.82, "max": 195.52, "n": 3 }, "p99_tpot_ms": { "median": 257.3, "min": 245.98, "max": 266.42, "n": 3 }, "p99_itl_ms": { "median": 1578.94, "min": 1530.64, "max": 1589.23, "n": 3 }, "median_itl_ms": { "median": 76.53, "min": 76.49, "max": 76.54, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.31, "min": 0.3, "max": 0.31, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 675.0, "min": 660.0, "max": 690.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.671 }, "ttft_quantiles_ms": [ 1898.1, 3195.3, 4503.0, 5839.7, 6067.0, 6173.7, 6282.6, 6676.4, 6750.6, 7104.9, 7152.6, 9589.5, 10287.5, 11469.5, 11520.6, 11631.8, 11660.0, 11675.1, 11697.7, 11751.8, 11804.1, 12686.2, 12885.9, 12984.7, 13204.4, 13349.2, 13553.5, 13742.3, 13800.5, 13911.6, 14141.7, 14660.9, 14861.3, 15704.3, 15916.1, 16049.1, 16491.7, 18270.4, 18957.0, 19359.5, 19501.2, 19624.8, 19770.1, 19992.8, 20409.9, 20507.3, 20565.4, 20710.8, 21066.4, 21468.1, 21500.3 ], "tpot_quantiles_ms": [ 73.4, 108.2, 140.2, 140.4, 140.7, 141.4, 141.6, 141.9, 142.0, 142.5, 142.7, 142.9, 143.5, 143.8, 145.2, 149.5, 149.9, 151.9, 153.5, 155.8, 158.1, 167.8, 176.5, 184.4, 190.7, 192.3, 195.7, 203.2, 209.6, 214.0, 215.2, 218.8, 220.7, 223.0, 225.2, 227.7, 229.1, 230.8, 231.2, 232.0, 234.3, 235.6, 237.0, 239.4, 240.9, 241.0, 241.9, 243.1, 244.3, 246.3, 271.0 ] } ] } }, "waves": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 156.93, "min": 145.48, "max": 175.74, "n": 3 }, "median_ttft_ms": { "median": 291.06, "min": 270.82, "max": 301.37, "n": 3 }, "p99_ttft_ms": { "median": 1169.39, "min": 980.57, "max": 1394.86, "n": 3 }, "median_tpot_ms": { "median": 87.38, "min": 78.54, "max": 91.58, "n": 3 }, "p99_tpot_ms": { "median": 123.63, "min": 104.87, "max": 147.47, "n": 3 }, "p99_itl_ms": { "median": 299.88, "min": 290.08, "max": 314.2, "n": 3 }, "median_itl_ms": { "median": 67.42, "min": 67.19, "max": 67.52, "n": 3 }, "goodput_rps": { "median": 2.14, "min": 1.96, "max": 2.32, "n": 3 }, "request_throughput": { "median": 2.39, "min": 2.2, "max": 2.5, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1402.5, "min": 1395.0, "max": 1417.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 93.1, 161.7, 163.1, 178.7, 187.4, 191.3, 191.6, 192.0, 192.4, 192.8, 199.0, 200.9, 202.1, 204.4, 206.1, 208.1, 217.3, 225.8, 238.5, 265.3, 270.4, 272.9, 273.4, 275.0, 276.9, 280.9, 284.7, 298.2, 300.4, 305.0, 306.8, 308.5, 311.2, 324.2, 329.9, 338.8, 373.6, 402.2, 467.5, 733.4, 734.7, 736.3, 803.3, 804.9, 904.1, 905.0, 906.5, 979.7, 1168.7, 1393.9, 1395.0 ], "tpot_quantiles_ms": [ 64.6, 66.5, 67.0, 67.6, 69.0, 69.7, 70.2, 72.2, 72.6, 73.7, 74.4, 74.8, 75.6, 76.5, 77.2, 77.9, 78.4, 78.6, 78.8, 79.0, 80.7, 80.8, 81.1, 82.3, 83.2, 84.8, 85.4, 85.6, 87.2, 88.0, 88.8, 89.4, 89.7, 89.9, 90.4, 91.3, 91.8, 92.2, 92.3, 92.6, 93.2, 93.4, 94.0, 94.9, 95.3, 96.6, 97.7, 98.5, 100.9, 112.2, 189.9 ] }, { "x": 0.5, "output_throughput": { "median": 154.15, "min": 150.66, "max": 156.09, "n": 3 }, "median_ttft_ms": { "median": 363.6, "min": 354.42, "max": 383.55, "n": 3 }, "p99_ttft_ms": { "median": 1772.14, "min": 1762.2, "max": 1916.34, "n": 3 }, "median_tpot_ms": { "median": 91.44, "min": 91.22, "max": 91.67, "n": 3 }, "p99_tpot_ms": { "median": 103.16, "min": 99.44, "max": 103.27, "n": 3 }, "p99_itl_ms": { "median": 395.67, "min": 393.22, "max": 425.85, "n": 3 }, "median_itl_ms": { "median": 68.64, "min": 68.63, "max": 68.71, "n": 3 }, "goodput_rps": { "median": 1.14, "min": 1.11, "max": 1.2, "n": 3 }, "request_throughput": { "median": 1.59, "min": 1.57, "max": 1.6, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1380.0, "min": 1335.0, "max": 1395.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.27 }, "ttft_quantiles_ms": [ 92.7, 195.8, 251.8, 270.4, 273.1, 278.7, 279.8, 282.2, 284.3, 290.1, 299.0, 304.4, 307.0, 308.8, 310.5, 312.2, 315.8, 317.1, 320.8, 323.9, 327.5, 335.5, 341.9, 349.2, 358.4, 364.3, 376.5, 378.3, 380.5, 387.7, 397.7, 402.6, 412.6, 436.8, 469.7, 494.2, 537.8, 580.0, 695.8, 1036.0, 1037.1, 1041.9, 1042.8, 1043.6, 1044.4, 1045.1, 1761.7, 1770.7, 1772.0, 1915.2, 1916.5 ], "tpot_quantiles_ms": [ 65.7, 68.4, 69.3, 71.7, 74.7, 76.4, 77.5, 79.4, 81.2, 82.5, 83.8, 84.3, 85.0, 86.1, 87.2, 87.6, 87.9, 88.1, 88.3, 88.8, 89.3, 89.9, 90.2, 90.6, 90.9, 91.3, 91.6, 91.8, 92.0, 92.3, 92.8, 93.0, 93.3, 93.8, 94.1, 94.1, 94.5, 94.7, 94.9, 95.2, 95.4, 95.8, 96.4, 96.6, 97.1, 97.3, 97.7, 98.9, 99.4, 101.8, 105.5 ] }, { "x": 1.0, "output_throughput": { "median": 167.06, "min": 162.86, "max": 180.29, "n": 3 }, "median_ttft_ms": { "median": 1995.7, "min": 1802.48, "max": 2131.29, "n": 3 }, "p99_ttft_ms": { "median": 5275.51, "min": 2584.64, "max": 5508.73, "n": 3 }, "median_tpot_ms": { "median": 77.28, "min": 75.71, "max": 78.04, "n": 3 }, "p99_tpot_ms": { "median": 103.25, "min": 88.41, "max": 104.76, "n": 3 }, "p99_itl_ms": { "median": 81.53, "min": 80.56, "max": 91.93, "n": 3 }, "median_itl_ms": { "median": 69.91, "min": 69.71, "max": 70.06, "n": 3 }, "goodput_rps": { "median": 0.26, "min": 0.24, "max": 0.29, "n": 3 }, "request_throughput": { "median": 1.31, "min": 1.27, "max": 1.41, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1350.0, "min": 1350.0, "max": 1395.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.301 }, "ttft_quantiles_ms": [ 89.9, 192.7, 199.2, 411.6, 507.4, 634.1, 652.4, 746.6, 795.8, 812.0, 888.3, 1275.9, 1277.0, 1551.9, 1685.6, 1691.0, 1692.8, 1703.2, 1705.0, 1830.2, 1831.3, 1910.2, 1911.8, 1913.1, 1995.0, 1996.2, 2033.6, 2034.8, 2131.9, 2295.1, 2296.4, 2330.5, 2411.2, 2463.5, 2465.1, 2508.7, 2511.0, 2515.8, 2544.3, 2545.8, 2584.8, 2883.2, 2885.0, 3381.6, 3891.6, 4031.5, 4033.3, 4496.7, 4498.9, 5275.4, 5508.9 ], "tpot_quantiles_ms": [ 69.7, 69.7, 69.7, 69.8, 69.9, 69.9, 69.9, 69.9, 69.9, 70.1, 70.1, 70.2, 70.6, 72.2, 72.2, 72.3, 72.3, 72.3, 74.5, 74.5, 75.2, 75.2, 76.2, 76.2, 76.4, 76.5, 76.8, 76.8, 77.3, 77.3, 77.7, 77.7, 78.0, 78.1, 80.3, 80.4, 81.0, 81.0, 81.7, 81.7, 81.8, 83.9, 85.1, 85.8, 87.3, 87.9, 90.8, 93.8, 99.7, 99.9, 113.4 ] } ] }, "fp8": { "kv_pool_tokens": 142369, "points": [ { "x": 0.0, "output_throughput": { "median": 281.37, "min": 256.35, "max": 301.66, "n": 3 }, "median_ttft_ms": { "median": 129.13, "min": 121.85, "max": 129.59, "n": 3 }, "p99_ttft_ms": { "median": 177.48, "min": 176.92, "max": 179.33, "n": 3 }, "median_tpot_ms": { "median": 47.98, "min": 47.29, "max": 49.77, "n": 3 }, "p99_tpot_ms": { "median": 56.84, "min": 54.45, "max": 62.14, "n": 3 }, "p99_itl_ms": { "median": 94.88, "min": 94.15, "max": 117.46, "n": 3 }, "median_itl_ms": { "median": 41.58, "min": 41.56, "max": 41.7, "n": 3 }, "goodput_rps": { "median": 4.1, "min": 3.95, "max": 4.4, "n": 3 }, "request_throughput": { "median": 4.1, "min": 3.95, "max": 4.4, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 2040.0, "min": 2017.5, "max": 2040.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 83.9, 90.1, 104.6, 109.3, 110.3, 111.4, 111.7, 112.2, 112.6, 113.2, 113.6, 114.1, 114.4, 115.2, 115.9, 117.7, 119.2, 119.5, 119.7, 120.6, 121.7, 123.3, 128.0, 128.1, 128.4, 128.5, 128.9, 129.1, 129.2, 129.4, 129.4, 129.6, 129.7, 129.7, 129.8, 130.4, 131.2, 132.4, 133.1, 134.6, 136.7, 136.9, 138.1, 139.8, 165.3, 172.1, 172.9, 174.1, 174.4, 178.3, 180.3 ], "tpot_quantiles_ms": [ 37.2, 40.3, 41.3, 41.8, 42.2, 43.2, 43.7, 44.1, 44.4, 44.6, 45.0, 45.0, 45.5, 45.8, 45.9, 46.1, 46.4, 46.8, 47.0, 47.1, 47.3, 47.5, 47.6, 47.7, 47.8, 47.9, 48.1, 48.2, 48.5, 48.8, 49.0, 49.2, 49.4, 49.7, 49.9, 50.3, 50.5, 51.1, 51.4, 51.6, 51.8, 52.0, 52.3, 52.6, 53.0, 53.2, 53.7, 54.0, 54.8, 55.9, 64.3 ] }, { "x": 0.5, "output_throughput": { "median": 296.68, "min": 290.96, "max": 303.27, "n": 3 }, "median_ttft_ms": { "median": 130.3, "min": 129.33, "max": 130.43, "n": 3 }, "p99_ttft_ms": { "median": 180.2, "min": 180.01, "max": 180.8, "n": 3 }, "median_tpot_ms": { "median": 47.43, "min": 47.4, "max": 48.0, "n": 3 }, "p99_tpot_ms": { "median": 50.96, "min": 50.76, "max": 51.25, "n": 3 }, "p99_itl_ms": { "median": 95.35, "min": 95.08, "max": 95.65, "n": 3 }, "median_itl_ms": { "median": 42.34, "min": 42.31, "max": 42.36, "n": 3 }, "goodput_rps": { "median": 3.06, "min": 3.06, "max": 3.07, "n": 3 }, "request_throughput": { "median": 3.06, "min": 3.06, "max": 3.07, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1395.0, "max": 1440.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.1, 99.3, 113.2, 114.6, 115.8, 116.4, 117.0, 117.5, 118.3, 118.6, 119.4, 120.2, 120.4, 120.9, 121.5, 122.0, 122.5, 123.3, 124.2, 125.2, 128.8, 129.4, 129.7, 129.9, 130.0, 130.2, 130.4, 130.4, 130.5, 130.5, 130.6, 130.7, 130.8, 130.9, 131.0, 131.2, 131.4, 131.7, 131.9, 132.5, 133.3, 134.3, 136.7, 137.8, 138.1, 138.4, 173.7, 175.3, 175.9, 179.1, 183.3 ], "tpot_quantiles_ms": [ 40.7, 41.5, 42.1, 42.2, 42.7, 43.2, 43.4, 43.8, 44.1, 44.6, 44.8, 45.1, 45.7, 45.8, 46.0, 46.1, 46.3, 46.4, 46.5, 46.7, 46.9, 47.1, 47.2, 47.3, 47.4, 47.5, 47.7, 47.8, 47.9, 48.1, 48.3, 48.3, 48.4, 48.6, 48.6, 48.7, 48.9, 49.0, 49.2, 49.2, 49.3, 49.4, 49.5, 49.6, 49.7, 49.8, 49.9, 50.2, 50.3, 50.8, 52.9 ] }, { "x": 1.0, "output_throughput": { "median": 270.14, "min": 267.65, "max": 271.17, "n": 3 }, "median_ttft_ms": { "median": 1525.62, "min": 1522.53, "max": 1584.47, "n": 3 }, "p99_ttft_ms": { "median": 2373.66, "min": 2279.2, "max": 2480.39, "n": 3 }, "median_tpot_ms": { "median": 47.16, "min": 47.05, "max": 47.79, "n": 3 }, "p99_tpot_ms": { "median": 59.44, "min": 58.7, "max": 59.78, "n": 3 }, "p99_itl_ms": { "median": 54.6, "min": 52.88, "max": 56.37, "n": 3 }, "median_itl_ms": { "median": 43.19, "min": 43.15, "max": 43.22, "n": 3 }, "goodput_rps": { "median": 0.43, "min": 0.42, "max": 0.43, "n": 3 }, "request_throughput": { "median": 2.11, "min": 2.09, "max": 2.12, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1185.0, "min": 1177.5, "max": 1185.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.038 }, "ttft_quantiles_ms": [ 83.4, 151.1, 156.5, 167.6, 509.4, 515.3, 520.2, 525.8, 529.0, 582.2, 709.0, 1063.7, 1064.8, 1068.7, 1070.4, 1082.6, 1084.1, 1398.5, 1400.3, 1513.9, 1515.7, 1520.9, 1521.5, 1522.4, 1523.2, 1525.4, 1527.0, 1583.0, 1584.1, 1711.5, 1713.0, 1727.3, 1729.5, 1876.4, 1906.6, 1907.7, 1921.2, 1922.6, 1928.9, 1929.8, 1985.1, 1994.2, 1996.3, 2000.6, 2057.9, 2070.1, 2085.8, 2087.9, 2278.9, 2373.6, 2480.8 ], "tpot_quantiles_ms": [ 43.1, 43.2, 43.2, 43.2, 43.2, 43.2, 43.3, 43.3, 43.3, 43.3, 43.3, 43.4, 43.4, 43.4, 43.4, 43.5, 46.1, 46.1, 47.0, 47.0, 47.0, 47.0, 47.0, 47.0, 47.2, 47.2, 47.8, 47.8, 47.8, 47.8, 47.8, 48.5, 48.5, 48.8, 48.8, 49.8, 49.8, 49.8, 49.9, 49.9, 49.9, 54.9, 54.9, 55.0, 56.7, 57.2, 57.3, 57.6, 58.0, 58.5, 61.7 ] } ] }, "sglang-h2h": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 158.18, "min": 146.7, "max": 176.92, "n": 3 }, "median_ttft_ms": { "median": 283.34, "min": 264.59, "max": 290.99, "n": 3 }, "p99_ttft_ms": { "median": 1134.48, "min": 946.31, "max": 1369.77, "n": 3 }, "median_tpot_ms": { "median": 86.71, "min": 78.14, "max": 90.6, "n": 3 }, "p99_tpot_ms": { "median": 119.93, "min": 103.64, "max": 145.11, "n": 3 }, "p99_itl_ms": { "median": 301.93, "min": 287.18, "max": 304.9, "n": 3 }, "median_itl_ms": { "median": 67.22, "min": 66.96, "max": 67.42, "n": 3 }, "goodput_rps": { "median": 2.2, "min": 1.94, "max": 2.33, "n": 3 }, "request_throughput": { "median": 2.41, "min": 2.22, "max": 2.52, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1485.0, "min": 1485.0, "max": 1500.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 92.2, 160.8, 162.3, 176.3, 187.3, 190.8, 191.2, 191.8, 192.0, 192.7, 196.3, 200.7, 202.0, 203.6, 205.5, 206.8, 217.3, 221.9, 234.7, 259.1, 261.8, 265.5, 266.6, 269.4, 271.0, 272.0, 273.6, 284.1, 293.0, 297.9, 299.7, 302.7, 307.9, 314.4, 323.3, 338.0, 365.1, 390.2, 453.1, 713.0, 714.2, 715.6, 778.6, 780.0, 873.5, 874.4, 876.0, 945.8, 1134.0, 1368.7, 1370.0 ], "tpot_quantiles_ms": [ 64.5, 66.4, 66.9, 67.4, 68.4, 69.5, 69.9, 72.0, 72.4, 73.5, 74.2, 74.5, 75.3, 76.3, 77.0, 77.7, 78.0, 78.2, 78.5, 78.8, 80.1, 80.4, 80.8, 81.5, 82.3, 84.1, 84.7, 84.9, 86.3, 87.1, 88.1, 88.4, 88.7, 88.9, 89.5, 90.4, 90.7, 91.0, 91.4, 91.9, 92.1, 92.3, 92.9, 93.5, 94.3, 95.4, 96.4, 97.4, 99.5, 110.8, 186.5 ] }, { "x": 0.5, "output_throughput": { "median": 159.55, "min": 155.75, "max": 162.03, "n": 3 }, "median_ttft_ms": { "median": 325.06, "min": 321.49, "max": 342.17, "n": 3 }, "p99_ttft_ms": { "median": 1714.26, "min": 1711.37, "max": 1865.8, "n": 3 }, "median_tpot_ms": { "median": 87.78, "min": 87.58, "max": 87.81, "n": 3 }, "p99_tpot_ms": { "median": 96.34, "min": 95.17, "max": 97.74, "n": 3 }, "p99_itl_ms": { "median": 362.58, "min": 360.23, "max": 370.02, "n": 3 }, "median_itl_ms": { "median": 68.35, "min": 68.33, "max": 68.39, "n": 3 }, "goodput_rps": { "median": 1.25, "min": 1.22, "max": 1.27, "n": 3 }, "request_throughput": { "median": 1.64, "min": 1.63, "max": 1.65, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1470.0, "min": 1432.5, "max": 1477.5, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 93.1, 194.5, 260.5, 263.9, 267.4, 268.6, 269.6, 270.8, 274.2, 275.6, 276.7, 284.6, 290.4, 293.1, 299.1, 300.7, 303.8, 304.7, 305.4, 307.3, 308.4, 310.3, 310.6, 311.8, 316.5, 326.1, 335.7, 345.4, 350.4, 360.4, 364.3, 371.9, 380.8, 390.0, 402.1, 417.8, 430.1, 471.8, 567.2, 1007.3, 1008.5, 1010.0, 1011.6, 1166.9, 1168.1, 1169.2, 1710.6, 1712.5, 1713.7, 1864.7, 1866.1 ], "tpot_quantiles_ms": [ 65.1, 67.7, 68.2, 71.3, 72.2, 74.2, 75.9, 77.1, 77.7, 80.1, 80.9, 82.7, 83.1, 83.4, 83.9, 84.3, 84.9, 85.3, 85.8, 86.2, 86.7, 86.9, 87.0, 87.2, 87.3, 87.6, 88.1, 88.9, 89.0, 89.4, 89.7, 90.0, 90.5, 90.6, 90.7, 90.8, 91.0, 91.3, 91.4, 91.6, 91.7, 92.2, 92.3, 92.8, 93.5, 93.6, 93.8, 94.5, 94.9, 95.7, 103.2 ] }, { "x": 1.0, "output_throughput": { "median": 182.28, "min": 181.49, "max": 182.7, "n": 3 }, "median_ttft_ms": { "median": 1775.03, "min": 1668.0, "max": 1830.12, "n": 3 }, "p99_ttft_ms": { "median": 2449.88, "min": 2420.12, "max": 2494.15, "n": 3 }, "median_tpot_ms": { "median": 74.65, "min": 74.32, "max": 75.52, "n": 3 }, "p99_tpot_ms": { "median": 87.57, "min": 87.08, "max": 87.65, "n": 3 }, "p99_itl_ms": { "median": 80.57, "min": 79.28, "max": 80.91, "n": 3 }, "median_itl_ms": { "median": 69.64, "min": 69.63, "max": 69.73, "n": 3 }, "goodput_rps": { "median": 0.36, "min": 0.22, "max": 0.45, "n": 3 }, "request_throughput": { "median": 1.42, "min": 1.42, "max": 1.43, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1440.0, "min": 1432.5, "max": 1470.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 93.6, 189.2, 191.3, 214.8, 493.8, 508.5, 610.5, 613.2, 623.5, 630.9, 757.8, 920.8, 921.8, 1183.5, 1642.0, 1643.5, 1658.9, 1660.8, 1662.5, 1663.7, 1671.1, 1673.3, 1761.1, 1762.3, 1785.7, 1786.7, 1787.4, 1788.7, 1789.5, 1794.7, 1797.0, 1829.9, 1831.6, 1921.6, 1922.9, 2074.1, 2075.4, 2196.8, 2315.1, 2316.0, 2339.4, 2352.5, 2361.8, 2388.7, 2408.5, 2410.0, 2420.2, 2450.0, 2468.3, 2492.5, 2494.3 ], "tpot_quantiles_ms": [ 69.6, 69.6, 69.6, 69.7, 69.7, 69.7, 69.7, 69.8, 69.8, 69.8, 69.8, 69.8, 69.9, 70.0, 71.0, 71.0, 73.0, 73.0, 73.0, 74.0, 74.0, 74.2, 74.2, 74.3, 74.3, 74.4, 74.4, 75.1, 75.1, 75.1, 75.1, 75.5, 75.5, 76.0, 76.0, 76.3, 76.3, 78.5, 78.5, 80.0, 82.2, 83.0, 83.4, 83.6, 84.1, 84.3, 84.8, 85.6, 86.8, 87.3, 88.0 ] } ] }, "vllm": { "kv_pool_tokens": 24096, "points": [ { "x": 0.0, "output_throughput": { "median": 174.44, "min": 164.06, "max": 195.01, "n": 3 }, "median_ttft_ms": { "median": 283.85, "min": 223.09, "max": 284.5, "n": 3 }, "p99_ttft_ms": { "median": 1162.78, "min": 980.33, "max": 1304.61, "n": 3 }, "median_tpot_ms": { "median": 74.96, "min": 68.44, "max": 75.64, "n": 3 }, "p99_tpot_ms": { "median": 116.09, "min": 103.13, "max": 150.94, "n": 3 }, "p99_itl_ms": { "median": 201.88, "min": 180.03, "max": 215.05, "n": 3 }, "median_itl_ms": { "median": 66.27, "min": 66.15, "max": 66.39, "n": 3 }, "goodput_rps": { "median": 2.42, "min": 2.1, "max": 2.61, "n": 3 }, "request_throughput": { "median": 2.65, "min": 2.45, "max": 2.81, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1492.5, "min": 1470.0, "max": 1507.5, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 92.4, 196.7, 203.8, 204.4, 204.6, 204.9, 205.3, 205.5, 205.9, 206.2, 206.6, 207.2, 207.7, 208.6, 209.1, 210.9, 212.6, 213.6, 214.4, 214.8, 216.6, 220.2, 225.2, 230.1, 269.8, 272.2, 278.9, 283.5, 286.7, 290.9, 306.4, 311.5, 313.8, 323.0, 333.8, 339.9, 348.1, 372.7, 392.1, 647.0, 648.0, 648.5, 649.0, 649.6, 650.2, 979.0, 980.2, 1161.5, 1162.9, 1252.3, 1391.9 ], "tpot_quantiles_ms": [ 63.5, 64.7, 65.4, 65.5, 65.8, 66.0, 66.3, 66.6, 66.7, 66.8, 67.0, 67.0, 67.1, 67.1, 67.2, 67.3, 67.3, 67.5, 67.8, 68.9, 69.3, 70.3, 70.9, 71.7, 72.4, 73.0, 73.1, 73.5, 74.0, 74.7, 75.0, 75.7, 76.4, 77.2, 77.4, 77.7, 78.1, 78.5, 79.0, 79.2, 80.0, 80.5, 81.1, 81.8, 82.5, 83.3, 84.9, 88.6, 92.6, 101.8, 215.7 ] }, { "x": 0.5, "output_throughput": { "median": 169.6, "min": 165.18, "max": 173.44, "n": 3 }, "median_ttft_ms": { "median": 352.6, "min": 347.59, "max": 376.59, "n": 3 }, "p99_ttft_ms": { "median": 1796.9, "min": 1682.52, "max": 1850.83, "n": 3 }, "median_tpot_ms": { "median": 81.41, "min": 81.01, "max": 82.48, "n": 3 }, "p99_tpot_ms": { "median": 88.78, "min": 87.63, "max": 90.05, "n": 3 }, "p99_itl_ms": { "median": 190.72, "min": 189.37, "max": 218.99, "n": 3 }, "median_itl_ms": { "median": 67.78, "min": 67.75, "max": 67.79, "n": 3 }, "goodput_rps": { "median": 1.46, "min": 1.44, "max": 1.5, "n": 3 }, "request_throughput": { "median": 1.75, "min": 1.74, "max": 1.76, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1425.0, "max": 1455.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.5, 267.7, 271.8, 274.0, 275.6, 277.2, 278.3, 279.0, 280.3, 281.7, 283.0, 283.5, 290.5, 302.0, 305.9, 310.1, 313.1, 313.9, 315.9, 317.0, 317.9, 319.3, 321.8, 322.3, 338.9, 350.9, 362.1, 378.7, 381.9, 391.6, 401.7, 412.1, 420.5, 425.4, 430.9, 434.4, 443.3, 460.2, 561.7, 650.6, 651.2, 651.7, 703.2, 1242.9, 1244.6, 1252.9, 1254.2, 1682.7, 1796.4, 1813.4, 1912.8 ], "tpot_quantiles_ms": [ 64.6, 66.7, 67.6, 69.7, 71.0, 72.2, 72.7, 73.5, 74.6, 75.6, 76.6, 77.1, 77.3, 77.7, 77.9, 78.4, 78.8, 79.0, 79.3, 79.6, 79.9, 80.3, 80.7, 80.9, 81.2, 81.5, 81.7, 81.9, 82.4, 82.6, 82.7, 82.8, 83.1, 83.4, 83.7, 83.9, 84.0, 84.0, 84.1, 84.2, 84.4, 84.7, 84.9, 85.2, 85.5, 85.7, 85.9, 86.5, 87.2, 89.9, 90.6 ] }, { "x": 1.0, "output_throughput": { "median": 181.71, "min": 181.68, "max": 182.82, "n": 3 }, "median_ttft_ms": { "median": 1499.34, "min": 1460.33, "max": 1532.05, "n": 3 }, "p99_ttft_ms": { "median": 2315.72, "min": 2304.78, "max": 2392.47, "n": 3 }, "median_tpot_ms": { "median": 76.76, "min": 75.85, "max": 77.16, "n": 3 }, "p99_tpot_ms": { "median": 86.18, "min": 85.62, "max": 86.19, "n": 3 }, "p99_itl_ms": { "median": 502.21, "min": 500.1, "max": 503.08, "n": 3 }, "median_itl_ms": { "median": 69.65, "min": 69.56, "max": 69.69, "n": 3 }, "goodput_rps": { "median": 0.35, "min": 0.33, "max": 0.38, "n": 3 }, "request_throughput": { "median": 1.42, "min": 1.42, "max": 1.43, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1410.0, "min": 1410.0, "max": 1410.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.0, 347.4, 354.6, 358.7, 629.3, 634.0, 652.9, 654.9, 775.5, 786.6, 788.9, 870.0, 982.7, 1054.3, 1056.5, 1212.7, 1215.8, 1244.7, 1245.7, 1285.6, 1298.0, 1311.1, 1350.0, 1397.7, 1436.3, 1462.3, 1569.1, 1580.3, 1588.3, 1590.6, 1594.8, 1601.6, 1602.2, 1604.5, 1614.2, 1637.1, 1645.1, 1682.4, 1685.0, 1703.0, 1728.4, 1798.0, 1813.7, 1815.4, 1876.1, 1916.6, 2104.5, 2186.0, 2304.4, 2315.3, 2392.7 ], "tpot_quantiles_ms": [ 69.5, 69.6, 70.7, 72.2, 73.2, 73.9, 74.0, 74.0, 74.0, 74.0, 74.1, 74.7, 74.8, 74.8, 74.9, 75.0, 75.0, 75.4, 75.6, 76.0, 76.1, 76.1, 76.3, 76.4, 76.4, 76.7, 76.8, 76.8, 77.2, 77.3, 77.5, 77.9, 78.3, 78.6, 79.0, 79.2, 79.9, 80.6, 80.7, 80.8, 80.9, 80.9, 82.4, 82.6, 82.6, 83.2, 83.7, 84.1, 84.9, 85.9, 86.8 ] } ] }, "mixed-chunk": { "kv_pool_tokens": 24094, "points": [ { "x": 0.0, "output_throughput": { "median": 171.89, "min": 160.31, "max": 192.09, "n": 3 }, "median_ttft_ms": { "median": 288.07, "min": 267.68, "max": 306.74, "n": 3 }, "p99_ttft_ms": { "median": 1199.53, "min": 950.46, "max": 1373.64, "n": 3 }, "median_tpot_ms": { "median": 76.76, "min": 69.66, "max": 79.67, "n": 3 }, "p99_tpot_ms": { "median": 108.81, "min": 92.84, "max": 139.99, "n": 3 }, "p99_itl_ms": { "median": 195.97, "min": 165.12, "max": 197.18, "n": 3 }, "median_itl_ms": { "median": 67.19, "min": 66.96, "max": 67.2, "n": 3 }, "goodput_rps": { "median": 2.49, "min": 2.19, "max": 2.53, "n": 3 }, "request_throughput": { "median": 2.61, "min": 2.41, "max": 2.75, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1507.5, "min": 1500.0, "max": 1530.0, "n": 3 }, "max_temp_c": 81.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 92.0, 186.6, 193.5, 197.8, 198.8, 199.1, 199.3, 199.6, 200.5, 203.0, 205.9, 207.0, 208.8, 210.4, 211.5, 212.4, 217.6, 220.8, 226.7, 261.7, 266.2, 270.7, 278.4, 280.4, 284.1, 286.1, 298.4, 302.0, 303.9, 306.4, 310.2, 312.5, 316.4, 336.7, 337.9, 349.1, 368.8, 381.4, 423.3, 711.5, 712.6, 714.2, 787.1, 788.7, 877.3, 878.3, 879.9, 950.2, 1199.0, 1372.4, 1373.9 ], "tpot_quantiles_ms": [ 63.7, 65.2, 65.9, 66.2, 66.5, 66.9, 67.1, 67.2, 67.3, 67.4, 67.6, 67.6, 67.8, 67.9, 68.4, 68.7, 69.2, 69.5, 69.6, 70.7, 71.1, 71.6, 72.5, 74.1, 74.9, 75.6, 76.2, 76.7, 76.8, 77.2, 77.7, 78.4, 78.8, 79.1, 79.5, 79.7, 79.9, 80.3, 80.6, 80.9, 81.3, 81.7, 81.9, 82.1, 82.8, 83.3, 83.5, 85.4, 92.1, 94.1, 183.0 ] }, { "x": 0.5, "output_throughput": { "median": 171.77, "min": 165.23, "max": 173.56, "n": 3 }, "median_ttft_ms": { "median": 343.46, "min": 336.12, "max": 370.44, "n": 3 }, "p99_ttft_ms": { "median": 1702.93, "min": 1639.36, "max": 1855.06, "n": 3 }, "median_tpot_ms": { "median": 80.84, "min": 80.82, "max": 81.49, "n": 3 }, "p99_tpot_ms": { "median": 88.51, "min": 87.79, "max": 88.71, "n": 3 }, "p99_itl_ms": { "median": 193.12, "min": 192.54, "max": 229.92, "n": 3 }, "median_itl_ms": { "median": 68.46, "min": 68.42, "max": 68.46, "n": 3 }, "goodput_rps": { "median": 1.34, "min": 1.33, "max": 1.36, "n": 3 }, "request_throughput": { "median": 1.75, "min": 1.74, "max": 1.78, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1477.5, "min": 1470.0, "max": 1485.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 94.5, 202.1, 263.3, 268.5, 272.7, 275.2, 276.6, 278.0, 278.9, 282.2, 286.2, 296.6, 297.2, 300.6, 301.6, 305.5, 307.9, 309.5, 311.4, 313.0, 314.6, 318.0, 323.8, 330.7, 339.6, 349.9, 367.4, 373.9, 377.0, 380.1, 391.4, 392.4, 402.9, 412.1, 415.8, 427.9, 442.7, 457.0, 527.5, 1007.9, 1008.6, 1009.2, 1010.2, 1010.7, 1018.8, 1020.0, 1638.5, 1701.5, 1702.8, 1854.4, 1855.3 ], "tpot_quantiles_ms": [ 65.0, 67.1, 68.2, 69.8, 70.4, 71.9, 72.7, 73.0, 74.2, 75.9, 76.1, 76.7, 77.2, 77.4, 77.8, 78.1, 78.6, 78.9, 79.0, 79.5, 79.6, 80.0, 80.5, 80.6, 80.8, 81.0, 81.2, 81.4, 81.4, 81.6, 81.7, 81.8, 82.1, 82.2, 82.5, 82.6, 82.9, 83.0, 83.2, 83.5, 83.7, 83.9, 84.1, 84.2, 84.3, 84.4, 84.6, 85.5, 86.1, 87.0, 91.6 ] }, { "x": 1.0, "output_throughput": { "median": 182.25, "min": 181.13, "max": 182.65, "n": 3 }, "median_ttft_ms": { "median": 1947.33, "min": 1851.84, "max": 2002.32, "n": 3 }, "p99_ttft_ms": { "median": 2325.9, "min": 2303.95, "max": 2543.78, "n": 3 }, "median_tpot_ms": { "median": 73.08, "min": 72.37, "max": 74.1, "n": 3 }, "p99_tpot_ms": { "median": 86.14, "min": 86.01, "max": 86.71, "n": 3 }, "p99_itl_ms": { "median": 203.24, "min": 83.19, "max": 348.54, "n": 3 }, "median_itl_ms": { "median": 69.65, "min": 69.62, "max": 69.68, "n": 3 }, "goodput_rps": { "median": 0.45, "min": 0.09, "max": 0.53, "n": 3 }, "request_throughput": { "median": 1.42, "min": 1.42, "max": 1.43, "n": 3 }, "completed": [ 64, 64, 64 ], "gpu": { "median_sm_mhz": { "median": 1425.0, "min": 1425.0, "max": 1455.0, "n": 3 }, "max_temp_c": 85.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 93.9, 318.2, 329.1, 360.0, 650.1, 792.2, 853.4, 861.3, 948.5, 974.1, 986.8, 989.0, 991.4, 1247.9, 1249.5, 1283.2, 1286.0, 1365.8, 1369.1, 1386.6, 1795.8, 1797.3, 1945.5, 1946.9, 1949.3, 1952.0, 1952.9, 1953.9, 1955.6, 2037.4, 2040.4, 2051.6, 2054.5, 2085.5, 2088.1, 2093.4, 2096.0, 2176.3, 2230.5, 2248.2, 2274.4, 2304.3, 2315.8, 2325.1, 2326.9, 2327.9, 2342.1, 2439.2, 2443.4, 2520.2, 2578.9 ], "tpot_quantiles_ms": [ 69.6, 69.6, 69.6, 69.6, 69.7, 69.7, 69.8, 69.8, 70.5, 70.7, 70.7, 70.7, 70.7, 70.9, 71.9, 71.9, 72.0, 72.0, 72.1, 72.1, 72.5, 72.5, 72.6, 72.6, 72.6, 72.6, 72.6, 73.1, 73.1, 73.2, 73.2, 77.5, 77.5, 78.1, 78.1, 78.7, 78.9, 78.9, 80.4, 80.4, 80.7, 80.8, 81.0, 81.1, 81.2, 81.2, 81.8, 81.8, 85.8, 86.1, 87.1 ] } ] } }, "working-set": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 4.0, "output_throughput": { "median": 82.39, "min": 80.7, "max": 84.79, "n": 3 }, "median_ttft_ms": { "median": 359.5, "min": 347.91, "max": 362.14, "n": 3 }, "p99_ttft_ms": { "median": 2837.81, "min": 2275.9, "max": 2850.26, "n": 3 }, "median_tpot_ms": { "median": 71.18, "min": 68.79, "max": 71.19, "n": 3 }, "p99_tpot_ms": { "median": 150.6, "min": 138.28, "max": 150.62, "n": 3 }, "p99_itl_ms": { "median": 163.93, "min": 75.26, "max": 175.23, "n": 3 }, "median_itl_ms": { "median": 68.84, "min": 68.58, "max": 68.94, "n": 3 }, "goodput_rps": { "median": 1.93, "min": 1.89, "max": 1.99, "n": 3 }, "request_throughput": { "median": 2.57, "min": 2.52, "max": 2.65, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1455.0, "min": 1440.0, "max": 1815.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 95.0, 104.0, 109.1, 113.4, 146.1, 164.9, 168.1, 186.0, 294.9, 297.7, 299.0, 299.1, 300.3, 301.7, 312.6, 314.4, 317.2, 328.9, 330.3, 332.4, 345.9, 346.4, 347.6, 349.2, 357.9, 358.6, 359.2, 359.7, 360.3, 360.6, 361.4, 361.5, 362.7, 363.8, 396.9, 402.5, 403.5, 404.3, 408.6, 422.6, 761.6, 767.9, 1368.5, 1981.0, 2274.6, 2275.6, 2835.8, 2837.0, 2838.0, 2849.4, 2850.4 ], "tpot_quantiles_ms": [ 67.7, 67.7, 67.7, 67.7, 68.0, 68.7, 68.7, 68.8, 68.8, 68.8, 68.8, 68.8, 68.8, 68.8, 68.9, 69.1, 69.1, 69.1, 69.1, 69.1, 69.1, 69.1, 69.2, 69.2, 69.2, 71.1, 71.1, 71.1, 71.2, 71.2, 71.2, 71.2, 71.3, 71.3, 72.0, 72.0, 72.4, 72.5, 72.7, 72.7, 73.3, 74.8, 75.6, 76.9, 77.0, 77.3, 97.0, 136.0, 136.1, 138.9, 157.2 ] }, { "x": 16.0, "output_throughput": { "median": 37.95, "min": 36.15, "max": 38.08, "n": 3 }, "median_ttft_ms": { "median": 3259.19, "min": 2654.72, "max": 4038.94, "n": 3 }, "p99_ttft_ms": { "median": 7443.19, "min": 6933.57, "max": 8670.03, "n": 3 }, "median_tpot_ms": { "median": 89.52, "min": 86.93, "max": 94.88, "n": 3 }, "p99_tpot_ms": { "median": 279.93, "min": 245.8, "max": 297.66, "n": 3 }, "p99_itl_ms": { "median": 1619.06, "min": 1573.84, "max": 2045.52, "n": 3 }, "median_itl_ms": { "median": 72.46, "min": 71.82, "max": 72.51, "n": 3 }, "goodput_rps": { "median": 0.06, "min": 0.0, "max": 0.1, "n": 3 }, "request_throughput": { "median": 1.19, "min": 1.13, "max": 1.19, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 787.5, "min": 645.0, "max": 975.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.708 }, "ttft_quantiles_ms": [ 103.9, 152.3, 167.6, 169.4, 315.6, 938.7, 945.9, 983.3, 1367.0, 1393.0, 1441.7, 1619.2, 1682.1, 1841.0, 1912.3, 2005.4, 2461.2, 2590.6, 2601.0, 2619.6, 2639.5, 2678.7, 2837.8, 2938.5, 3045.0, 3133.7, 3249.3, 3400.6, 3664.5, 3701.6, 3916.7, 4039.7, 4081.9, 4241.6, 4364.9, 4409.2, 4527.7, 4615.8, 4691.0, 4735.7, 4970.6, 5158.7, 5587.0, 5656.7, 5916.7, 6153.6, 6450.8, 6617.3, 6932.9, 7594.5, 8670.6 ], "tpot_quantiles_ms": [ 69.2, 70.7, 71.2, 71.4, 71.4, 71.5, 71.9, 72.0, 72.1, 72.4, 72.6, 72.6, 72.7, 72.7, 73.0, 73.4, 74.0, 74.2, 74.5, 74.6, 75.6, 81.1, 82.7, 83.4, 85.8, 87.0, 89.5, 92.9, 95.6, 97.2, 98.7, 105.7, 115.2, 119.8, 121.2, 123.3, 129.0, 136.8, 139.9, 141.4, 159.1, 161.7, 165.8, 174.6, 176.8, 188.1, 194.6, 219.5, 222.9, 254.6, 346.5 ] }, { "x": 32.0, "output_throughput": { "median": 25.72, "min": 24.96, "max": 26.29, "n": 3 }, "median_ttft_ms": { "median": 5448.53, "min": 5393.19, "max": 5473.37, "n": 3 }, "p99_ttft_ms": { "median": 10399.04, "min": 10289.0, "max": 10753.61, "n": 3 }, "median_tpot_ms": { "median": 130.54, "min": 127.1, "max": 132.12, "n": 3 }, "p99_tpot_ms": { "median": 359.6, "min": 358.51, "max": 362.74, "n": 3 }, "p99_itl_ms": { "median": 3343.28, "min": 3333.14, "max": 3467.53, "n": 3 }, "median_itl_ms": { "median": 73.37, "min": 73.14, "max": 73.45, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.8, "min": 0.78, "max": 0.82, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 465.0, "min": 465.0, "max": 465.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.896 }, "ttft_quantiles_ms": [ 104.1, 167.8, 170.6, 964.5, 1440.8, 1542.8, 1759.9, 2007.2, 2503.8, 2618.6, 2808.1, 2903.7, 3201.0, 3464.9, 3787.2, 3992.5, 4124.2, 4202.2, 4347.1, 4435.8, 4520.3, 4640.7, 4903.6, 5000.9, 5263.8, 5396.6, 5509.3, 5754.2, 6051.9, 6240.7, 6393.3, 6510.2, 6697.1, 6792.4, 6955.7, 6984.1, 7050.7, 7120.2, 7236.1, 7331.8, 7543.6, 7742.9, 7984.2, 8205.0, 8620.7, 8902.8, 9144.6, 9323.5, 9704.7, 10195.5, 12300.2 ], "tpot_quantiles_ms": [ 71.1, 71.8, 72.5, 72.9, 73.0, 73.1, 73.3, 73.7, 73.9, 74.1, 74.2, 74.3, 79.3, 91.7, 92.5, 93.3, 94.8, 97.5, 98.7, 100.1, 100.8, 102.7, 108.0, 113.4, 127.1, 130.6, 135.4, 136.4, 141.6, 144.9, 148.6, 156.5, 169.3, 172.9, 176.9, 182.2, 186.7, 189.8, 193.8, 206.9, 218.3, 220.7, 229.7, 242.0, 258.9, 268.8, 279.9, 297.1, 317.8, 345.3, 411.4 ] }, { "x": 64.0, "output_throughput": { "median": 21.37, "min": 21.3, "max": 21.92, "n": 3 }, "median_ttft_ms": { "median": 6821.04, "min": 6810.72, "max": 6957.53, "n": 3 }, "p99_ttft_ms": { "median": 12035.66, "min": 11656.49, "max": 12106.21, "n": 3 }, "median_tpot_ms": { "median": 150.27, "min": 144.69, "max": 154.83, "n": 3 }, "p99_tpot_ms": { "median": 402.03, "min": 378.96, "max": 405.39, "n": 3 }, "p99_itl_ms": { "median": 4302.83, "min": 4284.5, "max": 4685.79, "n": 3 }, "median_itl_ms": { "median": 73.71, "min": 73.65, "max": 73.72, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.67, "min": 0.67, "max": 0.69, "n": 3 }, "completed": [ 512, 512, 512 ], "gpu": { "median_sm_mhz": { "median": 450.0, "min": 450.0, "max": 465.0, "n": 3 }, "max_temp_c": 91.0, "thermal_share": 0.947 }, "ttft_quantiles_ms": [ 105.8, 777.2, 1423.6, 1510.5, 1581.9, 1803.0, 1957.4, 2399.5, 3205.2, 3653.9, 3931.2, 4095.1, 4233.1, 4317.8, 4380.7, 4467.1, 4693.1, 5223.8, 5585.9, 6134.0, 6392.8, 6534.2, 6633.7, 6694.6, 6768.4, 6882.2, 6956.5, 7034.0, 7100.7, 7202.0, 7413.8, 7767.9, 8075.5, 8290.5, 8496.8, 8707.5, 8865.5, 8961.3, 9067.2, 9180.9, 9317.4, 9418.4, 9616.7, 9743.7, 9830.6, 10046.7, 10345.6, 10757.9, 11048.0, 11542.7, 12246.1 ], "tpot_quantiles_ms": [ 71.5, 72.7, 72.9, 73.0, 73.2, 73.9, 74.0, 74.1, 74.1, 74.2, 74.3, 74.4, 93.7, 96.1, 97.7, 98.5, 100.1, 101.3, 110.1, 127.5, 132.5, 135.6, 137.4, 140.5, 144.0, 148.5, 159.6, 171.5, 174.1, 178.2, 182.0, 185.9, 191.7, 207.1, 210.3, 216.9, 221.9, 228.2, 234.4, 244.1, 254.2, 265.6, 276.4, 284.7, 300.1, 316.8, 328.8, 346.2, 362.8, 378.9, 425.3 ] } ] }, "hicache": { "kv_pool_tokens": 24094, "points": [ { "x": 4.0, "output_throughput": { "median": 81.64, "min": 80.18, "max": 84.53, "n": 3 }, "median_ttft_ms": { "median": 361.32, "min": 355.74, "max": 362.53, "n": 3 }, "p99_ttft_ms": { "median": 2896.97, "min": 2305.82, "max": 2925.83, "n": 3 }, "median_tpot_ms": { "median": 71.0, "min": 68.98, "max": 71.23, "n": 3 }, "p99_tpot_ms": { "median": 152.17, "min": 139.48, "max": 153.49, "n": 3 }, "p99_itl_ms": { "median": 159.41, "min": 99.72, "max": 160.33, "n": 3 }, "median_itl_ms": { "median": 68.85, "min": 68.68, "max": 68.93, "n": 3 }, "goodput_rps": { "median": 1.91, "min": 1.88, "max": 1.98, "n": 3 }, "request_throughput": { "median": 2.55, "min": 2.51, "max": 2.64, "n": 3 }, "completed": [ 32, 32, 32 ], "gpu": { "median_sm_mhz": { "median": 1440.0, "min": 1410.0, "max": 1440.0, "n": 3 }, "max_temp_c": 86.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 104.6, 106.6, 108.9, 109.2, 110.0, 165.6, 167.6, 185.0, 187.5, 299.3, 301.7, 302.0, 304.0, 305.5, 307.2, 309.9, 312.7, 315.1, 354.7, 356.0, 357.8, 358.9, 360.5, 361.0, 361.3, 361.6, 361.9, 362.1, 362.6, 362.8, 363.4, 363.4, 364.0, 364.3, 391.5, 397.8, 404.5, 405.5, 410.4, 417.6, 758.9, 778.4, 1373.8, 2014.4, 2304.5, 2305.5, 2895.6, 2896.6, 2924.0, 2925.1, 2926.0 ], "tpot_quantiles_ms": [ 67.8, 67.8, 67.9, 67.9, 68.1, 68.9, 68.9, 68.9, 69.0, 69.0, 69.0, 69.0, 69.0, 69.1, 69.1, 69.1, 69.2, 69.2, 69.2, 69.2, 69.2, 69.3, 69.3, 69.3, 70.8, 70.9, 71.1, 71.1, 71.2, 71.2, 71.2, 71.2, 71.2, 72.0, 72.0, 72.6, 72.6, 72.6, 72.6, 73.5, 74.7, 75.5, 76.8, 77.1, 77.2, 77.3, 98.5, 137.3, 137.9, 140.2, 160.0 ] }, { "x": 16.0, "output_throughput": { "median": 74.75, "min": 74.2, "max": 77.32, "n": 3 }, "median_ttft_ms": { "median": 552.55, "min": 548.93, "max": 569.44, "n": 3 }, "p99_ttft_ms": { "median": 3115.23, "min": 2853.14, "max": 3305.07, "n": 3 }, "median_tpot_ms": { "median": 77.14, "min": 74.02, "max": 77.29, "n": 3 }, "p99_tpot_ms": { "median": 159.32, "min": 137.8, "max": 161.31, "n": 3 }, "p99_itl_ms": { "median": 258.08, "min": 182.99, "max": 265.5, "n": 3 }, "median_itl_ms": { "median": 72.93, "min": 72.26, "max": 73.31, "n": 3 }, "goodput_rps": { "median": 1.61, "min": 1.34, "max": 1.94, "n": 3 }, "request_throughput": { "median": 2.34, "min": 2.32, "max": 2.42, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1222.5, "min": 1050.0, "max": 1230.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.521 }, "ttft_quantiles_ms": [ 105.4, 153.8, 167.4, 169.6, 170.8, 175.7, 202.7, 321.6, 331.9, 336.9, 352.7, 356.2, 362.1, 386.0, 390.3, 410.3, 438.0, 442.9, 472.6, 500.1, 509.5, 523.2, 527.1, 530.3, 564.0, 567.9, 572.3, 578.0, 582.6, 601.1, 605.3, 623.8, 631.6, 666.4, 699.1, 708.4, 729.4, 774.0, 943.6, 1047.8, 1207.0, 1342.6, 1415.5, 1552.9, 1891.0, 2169.0, 2418.2, 2646.1, 2930.2, 3285.1, 3584.9 ], "tpot_quantiles_ms": [ 69.2, 71.2, 71.3, 71.4, 71.8, 72.0, 72.5, 72.5, 72.6, 72.7, 72.7, 72.7, 73.0, 73.2, 73.5, 73.6, 73.7, 73.8, 73.9, 74.0, 74.5, 74.6, 74.9, 75.1, 76.0, 76.2, 76.3, 76.8, 77.1, 77.2, 78.4, 79.0, 79.4, 80.5, 80.6, 81.4, 81.5, 81.9, 82.1, 83.6, 84.5, 87.9, 89.1, 92.0, 97.2, 102.8, 107.9, 109.7, 122.7, 141.4, 184.7 ] }, { "x": 32.0, "output_throughput": { "median": 69.29, "min": 69.23, "max": 69.39, "n": 3 }, "median_ttft_ms": { "median": 817.13, "min": 766.88, "max": 903.41, "n": 3 }, "p99_ttft_ms": { "median": 2963.65, "min": 2952.63, "max": 3770.68, "n": 3 }, "median_tpot_ms": { "median": 89.34, "min": 79.93, "max": 89.94, "n": 3 }, "p99_tpot_ms": { "median": 133.77, "min": 131.72, "max": 162.91, "n": 3 }, "p99_itl_ms": { "median": 265.55, "min": 252.51, "max": 267.94, "n": 3 }, "median_itl_ms": { "median": 74.98, "min": 74.42, "max": 75.0, "n": 3 }, "goodput_rps": { "median": 1.21, "min": 1.12, "max": 1.37, "n": 3 }, "request_throughput": { "median": 2.17, "min": 2.16, "max": 2.17, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 765.0, "min": 630.0, "max": 780.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.731 }, "ttft_quantiles_ms": [ 105.5, 167.4, 170.0, 172.1, 206.6, 241.8, 282.3, 314.9, 327.7, 404.3, 434.7, 449.9, 489.1, 508.3, 528.4, 562.6, 583.7, 613.0, 625.3, 654.9, 670.8, 691.7, 735.3, 754.8, 767.7, 817.6, 825.2, 834.1, 842.8, 849.2, 854.1, 861.8, 866.6, 902.9, 919.5, 932.9, 940.0, 951.2, 969.3, 988.3, 1005.7, 1183.1, 1398.8, 1420.1, 1708.9, 1838.7, 1986.5, 2178.3, 2338.7, 2952.4, 3772.8 ], "tpot_quantiles_ms": [ 71.4, 71.9, 72.7, 73.2, 73.5, 73.7, 73.9, 74.1, 74.4, 74.5, 74.8, 75.1, 75.5, 76.4, 76.7, 77.6, 78.3, 78.6, 78.9, 79.4, 79.8, 80.5, 82.3, 84.8, 87.7, 88.4, 89.3, 89.7, 90.2, 90.4, 91.4, 91.5, 91.6, 94.1, 96.2, 96.9, 99.1, 99.8, 100.0, 100.7, 103.8, 104.9, 106.8, 108.8, 109.9, 111.2, 114.7, 121.4, 125.0, 131.1, 186.7 ] }, { "x": 64.0, "output_throughput": { "median": 35.93, "min": 34.11, "max": 36.44, "n": 3 }, "median_ttft_ms": { "median": 3609.78, "min": 3536.45, "max": 4234.57, "n": 3 }, "p99_ttft_ms": { "median": 8119.3, "min": 8014.3, "max": 8306.33, "n": 3 }, "median_tpot_ms": { "median": 89.78, "min": 79.27, "max": 92.96, "n": 3 }, "p99_tpot_ms": { "median": 289.74, "min": 280.0, "max": 294.81, "n": 3 }, "p99_itl_ms": { "median": 1644.99, "min": 1638.25, "max": 1744.22, "n": 3 }, "median_itl_ms": { "median": 73.89, "min": 73.82, "max": 74.07, "n": 3 }, "goodput_rps": { "median": 0.04, "min": 0.02, "max": 0.05, "n": 3 }, "request_throughput": { "median": 1.12, "min": 1.07, "max": 1.14, "n": 3 }, "completed": [ 512, 512, 512 ], "gpu": { "median_sm_mhz": { "median": 450.0, "min": 420.0, "max": 450.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.931 }, "ttft_quantiles_ms": [ 105.5, 168.9, 171.1, 174.6, 415.1, 466.9, 821.2, 1523.1, 1666.7, 1774.4, 1897.0, 1964.2, 2085.6, 2223.2, 2479.0, 2723.5, 2854.2, 2919.1, 2971.4, 2997.8, 3192.3, 3345.5, 3423.7, 3547.0, 3613.7, 3795.1, 3832.0, 3939.5, 4121.3, 4255.8, 4374.2, 4475.9, 4563.3, 4636.2, 4777.5, 4867.2, 4965.1, 5034.1, 5133.8, 5246.3, 5386.7, 5617.9, 5752.0, 6051.3, 6294.5, 6508.3, 6697.9, 6971.7, 7482.7, 7954.5, 10600.9 ], "tpot_quantiles_ms": [ 71.6, 72.6, 72.9, 73.0, 73.2, 73.5, 73.8, 74.0, 74.1, 74.1, 74.2, 74.2, 74.3, 74.4, 74.5, 74.5, 74.6, 74.6, 74.8, 75.3, 76.6, 78.6, 81.0, 87.7, 88.8, 89.9, 91.3, 92.5, 93.2, 93.7, 94.5, 95.2, 97.0, 99.4, 123.1, 131.0, 134.0, 137.3, 140.9, 142.7, 144.2, 148.2, 170.1, 176.0, 182.8, 185.8, 193.2, 227.6, 235.7, 273.5, 411.3 ] } ] } }, "mixed": { "default": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 44.09, "min": 44.06, "max": 44.88, "n": 3 }, "median_ttft_ms": { "median": 35999.18, "min": 9752.1, "max": 39340.7, "n": 3 }, "p99_ttft_ms": { "median": 68554.73, "min": 39544.53, "max": 77677.08, "n": 3 }, "median_tpot_ms": { "median": 150.09, "min": 147.25, "max": 154.43, "n": 3 }, "p99_tpot_ms": { "median": 192.27, "min": 189.2, "max": 207.56, "n": 3 }, "p99_itl_ms": { "median": 2578.99, "min": 2480.05, "max": 2863.96, "n": 3 }, "median_itl_ms": { "median": 76.59, "min": 76.4, "max": 76.6, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.01, "n": 3 }, "request_throughput": { "median": 0.28, "min": 0.26, "max": 0.3, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 772.5, "min": 690.0, "max": 772.5, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.761 }, "ttft_quantiles_ms": [ 77.9, 82.5, 525.1, 807.5, 1136.2, 1220.5, 1415.1, 1635.4, 1649.6, 2136.4, 2927.1, 4559.8, 6641.0, 7943.9, 8653.1, 9749.9, 11833.9, 13433.7, 14355.9, 15687.9, 18656.1, 20074.0, 20453.3, 21852.7, 23645.8, 24086.3, 28586.4, 30262.0, 31039.7, 31697.4, 33796.5, 35425.2, 36482.3, 37016.2, 38445.9, 39340.9, 40492.5, 41751.5, 43169.4, 44033.7, 45530.1, 49991.7, 51910.1, 53454.3, 54968.4, 58594.4, 59638.7, 60060.0, 69330.1, 71664.2, 77713.6 ], "tpot_quantiles_ms": [ 72.1, 72.4, 85.1, 94.0, 96.0, 101.0, 102.6, 105.5, 107.2, 111.8, 113.7, 116.8, 118.5, 123.1, 127.4, 127.9, 131.8, 132.6, 136.4, 139.0, 140.2, 141.6, 143.1, 147.6, 148.6, 149.1, 150.2, 152.8, 155.5, 155.7, 156.6, 157.5, 158.1, 159.2, 162.9, 163.8, 164.1, 166.0, 168.5, 170.7, 171.6, 173.7, 176.0, 176.8, 177.5, 180.7, 182.7, 187.0, 189.1, 195.1, 209.3 ] }, { "x": 1.0, "output_throughput": { "median": 45.44, "min": 45.01, "max": 46.71, "n": 3 }, "median_ttft_ms": { "median": 52367.56, "min": 29402.36, "max": 56992.61, "n": 3 }, "p99_ttft_ms": { "median": 109285.29, "min": 81340.01, "max": 121121.95, "n": 3 }, "median_tpot_ms": { "median": 149.84, "min": 149.82, "max": 154.71, "n": 3 }, "p99_tpot_ms": { "median": 189.04, "min": 185.37, "max": 214.39, "n": 3 }, "p99_itl_ms": { "median": 2751.61, "min": 2409.74, "max": 2985.09, "n": 3 }, "median_itl_ms": { "median": 76.53, "min": 76.49, "max": 76.6, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.29, "min": 0.26, "max": 0.31, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 825.0, "min": 750.0, "max": 900.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.755 }, "ttft_quantiles_ms": [ 77.6, 86.7, 525.6, 1167.6, 1321.5, 1802.6, 2049.9, 2472.3, 2608.8, 6732.2, 7308.6, 10690.2, 14980.3, 17426.8, 18584.0, 19798.6, 22077.0, 26090.7, 30665.1, 31469.7, 32358.6, 32998.1, 33653.2, 37640.8, 39750.9, 44928.3, 48953.3, 49415.7, 52637.3, 53076.2, 54242.0, 56775.9, 58617.5, 63663.8, 64630.3, 67496.0, 68801.4, 70662.4, 72138.9, 74137.9, 78614.5, 80880.9, 83839.6, 86345.8, 89361.0, 94251.0, 97477.6, 101102.8, 109306.9, 113448.3, 121375.4 ], "tpot_quantiles_ms": [ 72.1, 74.3, 90.1, 91.7, 96.8, 101.8, 103.7, 108.8, 112.8, 115.0, 118.2, 119.7, 122.9, 126.5, 128.7, 129.1, 135.1, 136.4, 139.9, 142.5, 144.5, 147.4, 148.6, 149.4, 150.0, 151.2, 152.9, 153.2, 154.0, 154.9, 155.4, 155.7, 156.8, 158.6, 159.8, 161.3, 162.3, 163.7, 165.3, 167.2, 168.3, 171.1, 172.3, 173.6, 174.6, 176.9, 180.2, 182.3, 187.2, 191.4, 219.3 ] }, { "x": 2.0, "output_throughput": { "median": 46.16, "min": 44.5, "max": 47.54, "n": 3 }, "median_ttft_ms": { "median": 59833.66, "min": 39476.3, "max": 67280.64, "n": 3 }, "p99_ttft_ms": { "median": 127864.35, "min": 103724.43, "max": 145708.79, "n": 3 }, "median_tpot_ms": { "median": 149.81, "min": 147.8, "max": 154.37, "n": 3 }, "p99_tpot_ms": { "median": 194.62, "min": 189.68, "max": 216.61, "n": 3 }, "p99_itl_ms": { "median": 2512.47, "min": 2332.1, "max": 3007.5, "n": 3 }, "median_itl_ms": { "median": 76.5, "min": 76.47, "max": 76.64, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.29, "min": 0.26, "max": 0.32, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 915.0, "min": 810.0, "max": 930.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.778 }, "ttft_quantiles_ms": [ 77.6, 82.3, 1196.4, 1813.9, 2197.0, 2580.4, 3292.6, 3924.9, 4033.1, 9831.4, 12347.7, 14488.5, 18793.7, 22553.6, 25014.8, 25297.0, 29069.9, 30793.6, 36563.7, 38387.8, 39120.5, 42007.6, 43139.3, 49348.2, 52844.1, 54880.6, 56231.2, 59698.7, 62591.9, 63385.1, 67054.4, 69134.3, 71029.4, 75740.1, 80600.2, 82774.6, 84086.6, 86179.9, 88185.5, 92129.8, 94847.6, 100525.5, 102091.0, 104213.3, 107204.0, 113772.4, 116264.2, 120526.1, 128010.8, 136609.3, 146637.7 ], "tpot_quantiles_ms": [ 73.4, 75.2, 84.6, 95.3, 99.9, 101.8, 103.3, 112.1, 114.3, 116.2, 117.0, 119.4, 121.4, 126.9, 129.9, 132.4, 133.2, 136.9, 139.7, 142.2, 144.0, 146.1, 146.5, 147.3, 149.0, 150.9, 151.5, 152.4, 153.1, 154.2, 154.8, 155.8, 156.1, 156.5, 157.5, 159.9, 161.2, 163.0, 163.8, 166.2, 168.3, 169.2, 171.0, 174.5, 178.3, 179.1, 180.8, 182.6, 188.2, 194.2, 218.8 ] } ] }, "chunk-1024": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 43.26, "min": 43.11, "max": 43.32, "n": 3 }, "median_ttft_ms": { "median": 37411.99, "min": 11475.16, "max": 41100.82, "n": 3 }, "p99_ttft_ms": { "median": 71914.41, "min": 44464.9, "max": 81568.95, "n": 3 }, "median_tpot_ms": { "median": 152.47, "min": 148.73, "max": 157.5, "n": 3 }, "p99_tpot_ms": { "median": 198.48, "min": 198.03, "max": 216.04, "n": 3 }, "p99_itl_ms": { "median": 2922.44, "min": 2766.91, "max": 3066.65, "n": 3 }, "median_itl_ms": { "median": 76.64, "min": 76.64, "max": 76.65, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.01, "n": 3 }, "request_throughput": { "median": 0.27, "min": 0.25, "max": 0.29, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 690.0, "min": 645.0, "max": 690.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.775 }, "ttft_quantiles_ms": [ 77.5, 82.9, 524.6, 1070.9, 1244.1, 1321.8, 1385.9, 1795.4, 2058.4, 2158.9, 3264.5, 6302.8, 7772.3, 8619.5, 10213.9, 11003.9, 13802.5, 15333.2, 16125.1, 18130.0, 19673.0, 20716.7, 21749.5, 24813.4, 25063.7, 26099.7, 29671.4, 32048.3, 34385.2, 35082.1, 36321.6, 37983.2, 38376.2, 39044.7, 41023.1, 41190.8, 42802.8, 43953.9, 45416.6, 46525.9, 47713.7, 52454.8, 55241.4, 56455.6, 58500.6, 61078.5, 62549.9, 64274.1, 72365.2, 74511.1, 81598.4 ], "tpot_quantiles_ms": [ 76.2, 77.3, 85.8, 93.8, 97.6, 99.8, 103.2, 104.8, 109.9, 113.0, 115.7, 117.3, 119.6, 120.7, 126.6, 131.8, 137.3, 139.1, 140.5, 143.5, 145.6, 146.9, 147.2, 149.8, 151.4, 153.1, 154.1, 156.5, 158.0, 159.0, 160.5, 161.5, 162.5, 164.8, 166.5, 167.3, 168.1, 173.0, 175.0, 176.1, 176.8, 178.3, 179.7, 180.4, 182.1, 185.4, 187.3, 192.1, 197.3, 203.4, 219.6 ] }, { "x": 1.0, "output_throughput": { "median": 44.78, "min": 43.92, "max": 45.46, "n": 3 }, "median_ttft_ms": { "median": 52859.49, "min": 29777.7, "max": 59333.3, "n": 3 }, "p99_ttft_ms": { "median": 110782.58, "min": 84724.35, "max": 124243.3, "n": 3 }, "median_tpot_ms": { "median": 152.09, "min": 147.93, "max": 155.89, "n": 3 }, "p99_tpot_ms": { "median": 200.42, "min": 194.39, "max": 218.13, "n": 3 }, "p99_itl_ms": { "median": 2843.09, "min": 2592.19, "max": 3096.66, "n": 3 }, "median_itl_ms": { "median": 76.53, "min": 76.49, "max": 76.65, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.28, "min": 0.26, "max": 0.31, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 907.5, "min": 840.0, "max": 915.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.771 }, "ttft_quantiles_ms": [ 77.8, 84.0, 555.0, 1232.2, 1429.7, 1818.5, 1894.1, 2549.9, 2825.9, 6865.9, 7570.5, 10839.6, 15137.0, 17600.2, 18764.5, 20301.1, 22211.9, 26610.2, 30612.7, 32021.8, 32701.3, 33570.6, 34014.7, 37950.8, 40844.1, 45864.1, 49371.1, 50398.1, 52935.4, 55318.9, 55570.9, 58617.9, 59545.5, 65838.4, 66727.4, 68115.3, 70067.0, 71991.0, 73024.2, 76135.4, 81297.2, 84131.9, 85249.8, 88124.3, 90980.8, 95997.2, 99053.6, 102550.7, 110944.7, 116408.5, 124250.2 ], "tpot_quantiles_ms": [ 73.4, 79.1, 89.0, 93.8, 97.8, 104.2, 105.0, 108.6, 113.9, 118.5, 120.2, 124.2, 125.6, 128.6, 130.3, 131.1, 134.4, 136.2, 140.8, 142.4, 144.8, 146.9, 148.1, 149.4, 151.0, 152.8, 154.6, 156.0, 156.9, 158.0, 159.0, 161.4, 162.0, 162.6, 164.4, 166.8, 167.1, 169.6, 170.7, 171.5, 171.8, 173.6, 176.8, 178.0, 180.9, 182.0, 182.9, 186.5, 194.9, 198.7, 223.3 ] }, { "x": 2.0, "output_throughput": { "median": 43.97, "min": 43.72, "max": 45.72, "n": 3 }, "median_ttft_ms": { "median": 62469.02, "min": 41871.45, "max": 69443.92, "n": 3 }, "p99_ttft_ms": { "median": 134142.52, "min": 109367.85, "max": 149207.84, "n": 3 }, "median_tpot_ms": { "median": 151.68, "min": 149.34, "max": 160.8, "n": 3 }, "p99_tpot_ms": { "median": 195.67, "min": 194.18, "max": 216.03, "n": 3 }, "p99_itl_ms": { "median": 2929.75, "min": 2454.91, "max": 3183.3, "n": 3 }, "median_itl_ms": { "median": 76.65, "min": 76.5, "max": 76.83, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.28, "min": 0.26, "max": 0.31, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 825.0, "min": 795.0, "max": 922.5, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.758 }, "ttft_quantiles_ms": [ 77.2, 83.7, 1231.4, 1969.8, 2272.7, 2615.8, 3127.9, 3781.2, 4181.9, 10082.3, 12546.4, 14552.5, 18826.4, 24039.8, 26154.7, 27106.2, 30603.1, 31517.1, 37801.2, 39896.1, 41069.8, 43242.1, 46209.7, 52328.8, 54526.2, 56643.8, 59120.3, 62280.4, 65642.0, 66001.8, 69874.2, 72249.8, 74991.0, 78003.0, 84719.9, 86025.9, 87358.4, 89376.0, 92650.2, 94503.4, 99856.8, 103591.3, 107416.8, 108271.5, 111496.5, 116658.7, 121014.5, 125335.0, 134322.0, 139589.8, 150105.0 ], "tpot_quantiles_ms": [ 73.5, 81.2, 89.0, 93.8, 99.4, 105.6, 106.7, 113.1, 116.0, 119.3, 121.2, 123.5, 126.1, 128.3, 133.1, 133.8, 136.7, 139.6, 140.5, 142.7, 145.6, 147.0, 149.8, 151.7, 153.5, 154.9, 156.5, 158.8, 159.0, 159.9, 161.7, 163.4, 164.1, 166.2, 167.4, 168.4, 169.2, 170.7, 172.0, 174.7, 175.2, 177.2, 178.9, 180.3, 180.6, 183.9, 186.1, 190.7, 195.3, 198.6, 223.5 ] } ] }, "default-second": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 46.51, "min": 43.84, "max": 56.28, "n": 3 }, "median_ttft_ms": { "median": 35724.4, "min": 1334.62, "max": 37911.66, "n": 3 }, "p99_ttft_ms": { "median": 69005.27, "min": 8165.59, "max": 70284.35, "n": 3 }, "median_tpot_ms": { "median": 142.49, "min": 107.45, "max": 155.71, "n": 3 }, "p99_tpot_ms": { "median": 182.08, "min": 123.35, "max": 212.48, "n": 3 }, "p99_itl_ms": { "median": 2521.36, "min": 1229.96, "max": 2550.99, "n": 3 }, "median_itl_ms": { "median": 76.22, "min": 74.85, "max": 76.5, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.02, "n": 3 }, "request_throughput": { "median": 0.28, "min": 0.27, "max": 0.38, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 1080.0, "min": 825.0, "max": 1425.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.759 }, "ttft_quantiles_ms": [ 79.0, 83.7, 499.8, 660.8, 779.0, 831.3, 1010.7, 1138.1, 1165.0, 1206.6, 1246.6, 1310.8, 1380.0, 1687.0, 2040.5, 2234.9, 2762.4, 2981.4, 3972.5, 4931.0, 6368.0, 7306.9, 7619.1, 8767.4, 14373.3, 16274.8, 17902.2, 19924.9, 22781.1, 25399.3, 26034.1, 27196.2, 33684.1, 35743.3, 36523.6, 38251.7, 38768.8, 39902.7, 40732.7, 43680.7, 44842.5, 46376.9, 49034.4, 51464.4, 52253.6, 54397.8, 57019.1, 61698.5, 63307.0, 68847.3, 71077.3 ], "tpot_quantiles_ms": [ 70.8, 75.5, 82.1, 83.7, 91.9, 93.6, 96.0, 97.8, 99.3, 101.6, 103.1, 105.6, 106.1, 108.8, 109.7, 110.3, 111.1, 111.8, 113.9, 114.1, 114.7, 116.7, 117.0, 118.6, 120.8, 122.2, 123.7, 132.3, 134.2, 136.7, 137.7, 140.1, 143.4, 145.8, 147.4, 149.1, 150.4, 153.2, 155.2, 157.5, 163.4, 165.3, 168.5, 171.3, 173.5, 174.3, 177.1, 180.7, 183.0, 191.4, 216.0 ] }, { "x": 1.0, "output_throughput": { "median": 46.57, "min": 45.6, "max": 46.68, "n": 3 }, "median_ttft_ms": { "median": 54011.86, "min": 29121.42, "max": 55076.59, "n": 3 }, "p99_ttft_ms": { "median": 108337.86, "min": 80497.06, "max": 113437.85, "n": 3 }, "median_tpot_ms": { "median": 147.58, "min": 141.7, "max": 151.66, "n": 3 }, "p99_tpot_ms": { "median": 191.47, "min": 181.6, "max": 202.05, "n": 3 }, "p99_itl_ms": { "median": 2490.94, "min": 2261.72, "max": 2680.65, "n": 3 }, "median_itl_ms": { "median": 76.6, "min": 76.22, "max": 76.81, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.29, "min": 0.27, "max": 0.31, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 952.5, "min": 900.0, "max": 1080.0, "n": 3 }, "max_temp_c": 90.0, "thermal_share": 0.776 }, "ttft_quantiles_ms": [ 79.0, 85.1, 513.2, 1191.1, 1343.7, 1773.7, 2028.7, 2484.0, 2604.9, 6702.5, 7332.6, 10612.4, 14386.2, 16344.5, 18101.5, 19111.7, 22096.1, 25405.8, 29738.5, 30875.7, 31609.0, 32188.3, 33606.6, 37927.8, 39252.7, 44884.4, 46286.4, 50832.8, 52015.4, 53798.9, 55203.6, 56582.2, 59907.9, 60815.0, 64312.0, 67221.0, 67561.9, 69020.0, 69680.5, 72466.0, 77845.7, 79350.8, 81490.5, 83535.0, 88267.8, 90686.9, 94189.3, 97827.2, 103424.5, 108698.2, 113470.2 ], "tpot_quantiles_ms": [ 72.7, 74.2, 91.2, 93.9, 97.6, 100.1, 103.9, 106.0, 109.9, 113.5, 115.5, 116.7, 119.8, 123.0, 128.4, 131.3, 133.3, 136.3, 137.7, 138.4, 140.5, 143.1, 144.3, 145.7, 146.8, 147.5, 147.7, 148.5, 149.7, 151.2, 152.9, 153.8, 154.8, 156.1, 157.0, 158.4, 159.1, 160.4, 164.4, 167.3, 167.5, 169.5, 171.5, 173.1, 173.7, 175.8, 176.0, 179.3, 182.4, 192.5, 204.2 ] }, { "x": 2.0, "output_throughput": { "median": 47.29, "min": 46.47, "max": 48.11, "n": 3 }, "median_ttft_ms": { "median": 58266.58, "min": 40568.31, "max": 64242.5, "n": 3 }, "p99_ttft_ms": { "median": 121859.13, "min": 105981.33, "max": 137558.62, "n": 3 }, "median_tpot_ms": { "median": 147.41, "min": 138.54, "max": 151.85, "n": 3 }, "p99_tpot_ms": { "median": 190.42, "min": 187.2, "max": 198.08, "n": 3 }, "p99_itl_ms": { "median": 2572.06, "min": 2364.33, "max": 2835.06, "n": 3 }, "median_itl_ms": { "median": 76.38, "min": 76.29, "max": 76.66, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.3, "min": 0.27, "max": 0.32, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 1095.0, "min": 915.0, "max": 1095.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.731 }, "ttft_quantiles_ms": [ 80.7, 83.9, 1198.7, 1999.4, 2222.7, 2575.6, 3270.2, 3916.2, 4009.1, 9944.6, 12235.4, 14406.7, 18174.4, 23125.0, 24663.3, 25714.3, 28735.6, 30135.7, 34865.5, 37158.6, 39207.4, 41750.2, 43542.2, 48446.0, 51756.7, 54511.3, 56157.2, 58131.8, 60768.4, 63919.3, 64565.6, 66953.0, 71826.7, 73939.5, 77581.8, 81450.0, 82027.9, 85033.1, 86684.2, 91056.6, 94379.4, 96147.8, 98963.0, 102639.4, 104576.0, 107516.0, 111290.3, 115158.6, 122017.1, 128892.9, 138131.7 ], "tpot_quantiles_ms": [ 69.3, 72.7, 83.3, 90.1, 94.7, 99.7, 102.8, 107.2, 108.8, 110.8, 114.9, 117.0, 118.8, 122.6, 126.8, 128.3, 132.4, 135.6, 135.9, 137.5, 138.3, 139.3, 142.9, 144.5, 146.0, 147.4, 148.2, 148.8, 150.4, 151.9, 153.0, 154.4, 154.8, 155.6, 157.0, 159.2, 159.7, 160.7, 161.3, 162.0, 164.3, 165.4, 165.9, 169.3, 170.6, 172.7, 173.3, 178.7, 183.1, 190.5, 202.1 ] } ] }, "mixed-chunk": { "kv_pool_tokens": 24094, "points": [ { "x": 0.5, "output_throughput": { "median": 46.54, "min": 44.49, "max": 47.29, "n": 3 }, "median_ttft_ms": { "median": 32187.17, "min": 9943.38, "max": 35756.72, "n": 3 }, "p99_ttft_ms": { "median": 60193.7, "min": 38773.38, "max": 68877.89, "n": 3 }, "median_tpot_ms": { "median": 143.25, "min": 141.5, "max": 144.28, "n": 3 }, "p99_tpot_ms": { "median": 182.36, "min": 175.56, "max": 199.39, "n": 3 }, "p99_itl_ms": { "median": 2405.97, "min": 2348.5, "max": 2447.21, "n": 3 }, "median_itl_ms": { "median": 76.37, "min": 76.14, "max": 76.4, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.01, "n": 3 }, "request_throughput": { "median": 0.3, "min": 0.27, "max": 0.3, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 1035.0, "min": 840.0, "max": 1050.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.75 }, "ttft_quantiles_ms": [ 78.7, 83.2, 532.0, 829.8, 1200.1, 1259.6, 1410.4, 1666.3, 1785.3, 2153.2, 2818.8, 4931.8, 7122.0, 8061.4, 8762.5, 9244.3, 11794.6, 13367.6, 14207.9, 15150.3, 16998.8, 17883.3, 19902.8, 21379.7, 22271.5, 23306.3, 25526.5, 27240.9, 29811.4, 30883.1, 31732.1, 32389.9, 33663.3, 34267.8, 35627.7, 36015.1, 36782.5, 38055.4, 38439.9, 39393.4, 40537.1, 43856.2, 46319.7, 47254.5, 48538.4, 51358.3, 51834.9, 52618.1, 60701.0, 62208.4, 69049.8 ], "tpot_quantiles_ms": [ 69.2, 71.3, 83.2, 92.3, 93.3, 95.4, 100.2, 102.4, 105.0, 108.8, 109.6, 110.6, 113.6, 116.2, 119.2, 124.8, 129.9, 131.0, 132.7, 133.3, 135.5, 137.1, 138.7, 140.5, 141.8, 143.3, 143.9, 144.6, 146.2, 146.8, 147.9, 149.9, 152.8, 155.5, 155.9, 157.7, 158.5, 160.6, 161.6, 163.0, 164.1, 165.5, 168.1, 168.7, 169.9, 170.6, 173.0, 175.0, 177.7, 183.5, 200.1 ] }, { "x": 1.0, "output_throughput": { "median": 47.48, "min": 47.45, "max": 47.81, "n": 3 }, "median_ttft_ms": { "median": 49943.6, "min": 27409.14, "max": 54438.48, "n": 3 }, "p99_ttft_ms": { "median": 100256.25, "min": 78304.85, "max": 112695.11, "n": 3 }, "median_tpot_ms": { "median": 144.89, "min": 140.68, "max": 147.96, "n": 3 }, "p99_tpot_ms": { "median": 183.34, "min": 178.03, "max": 193.05, "n": 3 }, "p99_itl_ms": { "median": 2289.3, "min": 2236.48, "max": 2430.97, "n": 3 }, "median_itl_ms": { "median": 76.39, "min": 76.28, "max": 76.61, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.3, "min": 0.28, "max": 0.32, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 1020.0, "min": 772.5, "max": 1050.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.75 }, "ttft_quantiles_ms": [ 78.1, 84.0, 539.5, 1200.8, 1338.3, 1847.7, 2128.0, 2568.5, 2696.9, 6994.4, 7281.0, 10506.9, 13941.3, 16026.0, 17736.3, 18387.6, 21444.9, 24872.6, 28050.7, 29508.6, 30542.1, 31109.7, 31821.1, 35344.4, 38020.3, 42528.3, 46162.5, 46392.2, 50213.5, 51164.4, 51593.3, 54166.7, 55460.4, 61264.8, 61946.7, 63172.4, 65366.7, 66616.9, 68192.5, 69645.4, 74328.9, 77044.6, 78440.4, 80326.4, 82874.2, 87686.3, 90023.0, 92871.4, 100616.8, 105760.2, 112880.7 ], "tpot_quantiles_ms": [ 73.3, 75.8, 87.1, 91.4, 94.9, 98.6, 104.4, 106.2, 108.9, 110.8, 115.0, 116.5, 120.2, 122.3, 123.0, 127.5, 128.5, 131.0, 133.5, 136.6, 138.1, 140.0, 140.4, 141.5, 143.2, 145.2, 146.2, 147.6, 147.8, 148.7, 149.6, 149.8, 151.4, 152.1, 152.9, 153.7, 154.0, 156.1, 158.3, 159.0, 160.6, 162.5, 162.8, 166.2, 167.5, 169.2, 170.3, 173.0, 179.0, 183.0, 199.1 ] }, { "x": 2.0, "output_throughput": { "median": 47.63, "min": 46.93, "max": 49.53, "n": 3 }, "median_ttft_ms": { "median": 57939.29, "min": 38107.85, "max": 63616.08, "n": 3 }, "p99_ttft_ms": { "median": 122724.5, "min": 99072.59, "max": 136833.4, "n": 3 }, "median_tpot_ms": { "median": 143.67, "min": 140.08, "max": 145.52, "n": 3 }, "p99_tpot_ms": { "median": 180.77, "min": 179.42, "max": 202.92, "n": 3 }, "p99_itl_ms": { "median": 2417.97, "min": 2063.73, "max": 2467.85, "n": 3 }, "median_itl_ms": { "median": 76.37, "min": 76.28, "max": 76.38, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 0.3, "min": 0.28, "max": 0.33, "n": 3 }, "completed": [ 48, 48, 48 ], "gpu": { "median_sm_mhz": { "median": 1042.5, "min": 967.5, "max": 1065.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.745 }, "ttft_quantiles_ms": [ 80.5, 92.6, 1208.8, 1800.6, 2277.1, 2649.1, 3314.4, 4018.5, 4056.4, 9802.5, 12070.0, 14370.2, 18819.2, 21725.5, 24078.4, 24590.8, 28030.9, 29579.3, 34342.1, 36139.6, 37425.2, 39347.6, 42724.6, 47914.8, 50079.3, 51846.9, 54186.6, 57804.0, 60437.0, 60910.1, 64317.5, 65853.6, 68315.6, 71416.8, 77416.0, 78616.7, 79909.8, 81486.6, 84705.9, 85957.8, 91178.4, 95042.5, 97530.7, 98625.5, 102160.6, 106990.1, 111439.5, 115151.1, 122878.5, 128212.0, 137324.1 ], "tpot_quantiles_ms": [ 71.3, 72.8, 83.9, 87.7, 93.7, 96.9, 102.7, 106.3, 109.2, 111.1, 113.7, 114.3, 116.0, 119.7, 125.1, 128.5, 129.9, 131.4, 133.9, 134.3, 135.7, 137.0, 140.3, 142.2, 143.5, 144.0, 144.4, 144.9, 145.9, 147.1, 147.3, 148.8, 149.3, 150.7, 151.1, 152.3, 153.1, 154.3, 154.5, 156.0, 159.2, 161.7, 163.3, 165.0, 166.6, 168.6, 172.0, 174.0, 177.9, 183.3, 204.6 ] } ] } } }