{ "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": { "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 ] } ] }, "sglang": { "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-b": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 13.49, "min": 13.09, "max": 13.79, "n": 3 }, "median_ttft_ms": { "median": 191.28, "min": 190.87, "max": 195.99, "n": 3 }, "p99_ttft_ms": { "median": 695.78, "min": 687.07, "max": 706.25, "n": 3 }, "median_tpot_ms": { "median": 60.36, "min": 60.34, "max": 60.36, "n": 3 }, "p99_tpot_ms": { "median": 104.17, "min": 104.14, "max": 104.17, "n": 3 }, "p99_itl_ms": { "median": 104.53, "min": 104.4, "max": 104.56, "n": 3 }, "median_itl_ms": { "median": 60.52, "min": 60.41, "max": 60.56, "n": 3 }, "goodput_rps": { "median": 0.09, "min": 0.07, "max": 0.09, "n": 3 }, "request_throughput": { "median": 0.11, "min": 0.1, "max": 0.11, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1215.0, "max": 1290.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.483 }, "ttft_quantiles_ms": [ 76.1, 76.1, 77.3, 78.0, 168.1, 173.2, 173.6, 180.9, 181.1, 182.5, 182.6, 183.0, 183.7, 184.0, 186.0, 186.1, 186.2, 186.2, 186.5, 187.4, 187.7, 188.4, 189.1, 189.5, 190.3, 190.9, 191.5, 191.7, 193.1, 202.9, 205.5, 217.8, 237.0, 315.8, 315.8, 405.9, 439.7, 445.6, 452.8, 496.4, 508.9, 563.0, 571.3, 575.1, 578.1, 641.4, 704.3, 705.4, 705.4, 706.3, 706.4 ], "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, 60.3, 60.4, 60.4, 60.4, 60.4, 60.6, 60.9, 63.0, 68.3, 68.3, 74.8, 75.3, 85.6, 85.7, 87.0, 87.8, 103.7, 103.8, 104.1, 104.1, 104.1, 104.2, 104.2, 104.2, 104.2, 104.2 ] }, { "x": 2.0, "output_throughput": { "median": 31.26, "min": 31.26, "max": 31.26, "n": 3 }, "median_ttft_ms": { "median": 132.2, "min": 132.08, "max": 132.55, "n": 3 }, "p99_ttft_ms": { "median": 147.43, "min": 142.43, "max": 149.08, "n": 3 }, "median_tpot_ms": { "median": 63.43, "min": 63.42, "max": 63.43, "n": 3 }, "p99_tpot_ms": { "median": 63.54, "min": 63.53, "max": 63.55, "n": 3 }, "p99_itl_ms": { "median": 64.27, "min": 64.27, "max": 64.29, "n": 3 }, "median_itl_ms": { "median": 63.41, "min": 63.41, "max": 63.42, "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": 1500.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 78.0, 78.0, 80.0, 80.2, 128.3, 128.5, 128.6, 128.7, 128.8, 129.0, 129.0, 129.0, 129.1, 129.2, 129.2, 129.3, 129.3, 129.3, 129.3, 129.3, 129.3, 129.5, 129.6, 129.6, 130.4, 130.6, 133.6, 134.6, 134.7, 134.8, 134.9, 135.0, 135.0, 135.0, 135.0, 139.1, 139.3, 139.5, 139.6, 139.7, 139.7, 140.0, 140.0, 140.2, 140.3, 141.0, 141.1, 141.2, 142.7, 148.6, 150.5 ], "tpot_quantiles_ms": [ 63.3, 63.3, 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, 63.4, 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 ] }, { "x": 4.0, "output_throughput": { "median": 61.79, "min": 61.79, "max": 61.8, "n": 3 }, "median_ttft_ms": { "median": 141.61, "min": 139.85, "max": 142.08, "n": 3 }, "p99_ttft_ms": { "median": 154.73, "min": 153.3, "max": 155.71, "n": 3 }, "median_tpot_ms": { "median": 64.1, "min": 64.09, "max": 64.1, "n": 3 }, "p99_tpot_ms": { "median": 64.2, "min": 64.19, "max": 64.2, "n": 3 }, "p99_itl_ms": { "median": 65.17, "min": 64.84, "max": 65.81, "n": 3 }, "median_itl_ms": { "median": 64.1, "min": 64.09, "max": 64.1, "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": 1500.0, "max": 1560.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 81.6, 81.6, 81.8, 84.1, 131.2, 131.9, 132.1, 132.3, 132.5, 132.7, 133.2, 133.5, 134.2, 139.0, 139.1, 139.6, 139.6, 139.6, 140.1, 140.1, 140.1, 140.2, 140.6, 140.6, 140.7, 141.0, 141.2, 141.2, 141.3, 141.6, 141.6, 141.8, 142.1, 142.1, 142.1, 142.3, 142.5, 142.9, 143.0, 143.4, 143.9, 144.4, 152.4, 153.0, 153.4, 153.9, 154.5, 154.8, 154.8, 155.3, 155.8 ], "tpot_quantiles_ms": [ 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, 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, 64.2 ] }, { "x": 8.0, "output_throughput": { "median": 112.79, "min": 112.67, "max": 112.82, "n": 3 }, "median_ttft_ms": { "median": 252.88, "min": 251.67, "max": 258.62, "n": 3 }, "p99_ttft_ms": { "median": 1320.06, "min": 1311.34, "max": 1331.49, "n": 3 }, "median_tpot_ms": { "median": 65.75, "min": 65.64, "max": 65.76, "n": 3 }, "p99_tpot_ms": { "median": 73.31, "min": 73.28, "max": 73.72, "n": 3 }, "p99_itl_ms": { "median": 67.12, "min": 66.81, "max": 67.81, "n": 3 }, "median_itl_ms": { "median": 65.69, "min": 65.66, "max": 65.76, "n": 3 }, "goodput_rps": { "median": 0.66, "min": 0.66, "max": 0.69, "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": 1455.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.0, 87.0, 134.1, 134.7, 138.2, 139.1, 139.7, 140.8, 161.1, 161.6, 162.2, 162.6, 162.9, 163.2, 163.5, 163.8, 164.2, 164.5, 164.8, 211.3, 211.7, 211.9, 212.2, 212.8, 213.4, 214.4, 291.4, 310.5, 335.0, 583.6, 592.3, 699.3, 700.5, 702.4, 703.4, 704.6, 855.0, 982.9, 984.0, 1101.4, 1113.8, 1114.4, 1217.9, 1218.8, 1225.8, 1226.4, 1227.2, 1286.3, 1292.4, 1311.5, 1361.8 ], "tpot_quantiles_ms": [ 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, 65.6, 65.6, 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, 66.0, 66.6, 67.0, 67.0, 67.7, 67.9, 69.3, 69.3, 69.9, 69.9, 69.9, 70.8, 70.8, 71.0, 71.1, 73.0, 73.4, 74.0 ] }, { "x": 16.0, "output_throughput": { "median": 201.3, "min": 200.87, "max": 201.33, "n": 3 }, "median_ttft_ms": { "median": 288.25, "min": 280.31, "max": 289.93, "n": 3 }, "p99_ttft_ms": { "median": 2487.38, "min": 2485.68, "max": 2500.83, "n": 3 }, "median_tpot_ms": { "median": 69.49, "min": 69.45, "max": 69.68, "n": 3 }, "p99_tpot_ms": { "median": 85.6, "min": 85.47, "max": 85.77, "n": 3 }, "p99_itl_ms": { "median": 84.81, "min": 80.64, "max": 88.74, "n": 3 }, "median_itl_ms": { "median": 69.5, "min": 69.4, "max": 69.53, "n": 3 }, "goodput_rps": { "median": 1.01, "min": 1.01, "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": 1455.0, "min": 1425.0, "max": 1455.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 96.5, 183.6, 184.8, 186.3, 189.4, 191.1, 192.6, 193.9, 195.1, 196.3, 218.8, 219.5, 220.2, 221.0, 221.5, 222.7, 223.8, 225.0, 240.9, 241.7, 242.8, 244.5, 262.3, 263.2, 264.8, 265.9, 351.2, 834.2, 837.5, 842.2, 877.1, 879.0, 896.4, 1305.1, 1306.2, 1371.2, 1422.9, 1445.1, 1689.2, 1691.3, 1696.9, 1698.4, 1706.4, 1774.3, 1984.1, 2013.3, 2027.4, 2296.2, 2486.2, 2499.7, 2519.1 ], "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, 69.4, 69.4, 69.4, 69.4, 69.5, 69.5, 69.5, 69.5, 69.5, 69.5, 69.6, 69.6, 69.8, 69.9, 73.7, 73.8, 74.2, 74.2, 74.2, 74.2, 74.9, 74.9, 75.0, 77.2, 77.3, 77.3, 77.9, 78.4, 81.2, 81.3, 81.3, 81.4, 81.5, 81.7, 85.5, 86.2 ] }, { "x": 32.0, "output_throughput": { "median": 275.14, "min": 274.84, "max": 275.15, "n": 3 }, "median_ttft_ms": { "median": 1800.22, "min": 1763.26, "max": 1819.41, "n": 3 }, "p99_ttft_ms": { "median": 4821.25, "min": 4772.07, "max": 4847.92, "n": 3 }, "median_tpot_ms": { "median": 101.79, "min": 101.73, "max": 102.2, "n": 3 }, "p99_tpot_ms": { "median": 113.44, "min": 113.3, "max": 113.73, "n": 3 }, "p99_itl_ms": { "median": 613.31, "min": 608.38, "max": 613.67, "n": 3 }, "median_itl_ms": { "median": 80.53, "min": 80.5, "max": 80.59, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.15, "min": 2.15, "max": 2.15, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1267.5, "max": 1275.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 123.5, 393.7, 527.1, 676.7, 823.4, 836.1, 934.0, 1250.8, 1276.2, 1282.2, 1386.2, 1421.9, 1454.4, 1566.7, 1611.8, 1634.4, 1667.9, 1723.0, 1729.0, 1732.0, 1737.0, 1758.2, 1772.1, 1781.1, 1796.7, 1799.4, 1802.0, 1820.1, 1827.6, 1829.6, 1831.4, 1834.7, 1838.0, 1849.2, 1863.2, 1867.3, 1955.9, 2102.0, 2284.9, 2337.6, 2396.2, 2422.2, 2441.5, 2474.9, 3011.1, 3081.9, 3595.3, 3836.2, 4304.6, 4772.1, 4848.2 ], "tpot_quantiles_ms": [ 80.1, 80.3, 82.6, 86.0, 87.1, 91.4, 94.7, 95.6, 99.9, 100.8, 100.8, 100.9, 100.9, 100.9, 101.1, 101.2, 101.4, 101.5, 101.5, 101.6, 101.6, 101.6, 101.7, 101.7, 101.7, 101.8, 101.9, 102.0, 102.2, 102.3, 102.3, 102.4, 102.8, 103.0, 103.3, 103.5, 103.6, 103.9, 104.5, 104.6, 104.9, 105.4, 106.7, 107.7, 108.6, 108.8, 109.5, 110.6, 111.2, 112.8, 114.6 ] }, { "x": 64.0, "output_throughput": { "median": 242.32, "min": 242.04, "max": 243.77, "n": 3 }, "median_ttft_ms": { "median": 11180.53, "min": 10245.98, "max": 11319.61, "n": 3 }, "p99_ttft_ms": { "median": 26201.73, "min": 25347.15, "max": 26522.34, "n": 3 }, "median_tpot_ms": { "median": 129.15, "min": 128.08, "max": 129.19, "n": 3 }, "p99_tpot_ms": { "median": 261.51, "min": 259.88, "max": 263.02, "n": 3 }, "p99_itl_ms": { "median": 1232.94, "min": 1187.45, "max": 1436.99, "n": 3 }, "median_itl_ms": { "median": 83.75, "min": 83.63, "max": 83.76, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.89, "min": 1.89, "max": 1.9, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 1005.0, "min": 1005.0, "max": 1005.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.443 }, "ttft_quantiles_ms": [ 174.1, 1316.4, 1901.1, 2538.6, 3648.2, 4278.4, 4874.5, 4974.8, 5029.9, 5114.1, 5436.4, 5512.6, 5539.4, 5589.8, 5606.1, 5718.8, 5812.4, 5898.5, 6087.5, 6674.0, 7514.2, 8720.9, 9381.3, 9872.1, 10233.5, 10668.4, 11251.0, 11539.0, 12472.2, 13780.9, 15932.5, 16049.5, 16507.8, 16552.6, 16578.7, 16662.8, 16707.9, 16905.7, 17930.5, 18484.4, 18935.8, 19254.1, 19732.5, 20322.9, 20959.2, 23309.6, 24020.0, 24108.9, 24728.8, 25410.5, 26675.9 ], "tpot_quantiles_ms": [ 71.8, 79.3, 119.9, 120.3, 120.4, 120.5, 120.6, 120.7, 120.9, 121.0, 121.2, 121.3, 121.4, 121.5, 121.6, 122.0, 122.2, 122.2, 122.3, 122.6, 122.7, 123.6, 124.6, 125.3, 126.7, 128.4, 131.2, 136.0, 140.2, 144.3, 148.8, 150.8, 155.3, 161.0, 164.0, 167.3, 170.2, 171.1, 173.9, 175.0, 176.9, 178.7, 180.4, 181.6, 184.3, 187.4, 191.3, 193.4, 206.2, 242.9, 304.5 ] } ] }, "vllm-eager": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 13.15, "min": 13.06, "max": 13.36, "n": 3 }, "median_ttft_ms": { "median": 190.05, "min": 185.96, "max": 194.92, "n": 3 }, "p99_ttft_ms": { "median": 683.52, "min": 641.67, "max": 684.15, "n": 3 }, "median_tpot_ms": { "median": 61.07, "min": 61.06, "max": 61.07, "n": 3 }, "p99_tpot_ms": { "median": 106.06, "min": 105.92, "max": 106.47, "n": 3 }, "p99_itl_ms": { "median": 106.76, "min": 106.75, "max": 106.97, "n": 3 }, "median_itl_ms": { "median": 61.21, "min": 61.2, "max": 61.21, "n": 3 }, "goodput_rps": { "median": 0.07, "min": 0.07, "max": 0.08, "n": 3 }, "request_throughput": { "median": 0.1, "min": 0.1, "max": 0.1, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1290.0, "min": 1260.0, "max": 1305.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.48 }, "ttft_quantiles_ms": [ 90.0, 90.0, 91.1, 91.8, 164.4, 168.4, 169.7, 175.3, 178.6, 181.5, 182.6, 183.1, 183.2, 183.6, 184.1, 184.6, 184.7, 184.7, 184.9, 185.5, 186.4, 187.0, 187.5, 188.1, 188.4, 188.6, 191.3, 191.7, 192.9, 198.6, 203.3, 215.3, 370.3, 441.7, 441.7, 455.6, 460.3, 536.7, 540.3, 542.3, 547.4, 548.1, 555.2, 557.8, 558.4, 613.1, 617.4, 628.2, 644.0, 705.6, 706.9 ], "tpot_quantiles_ms": [ 60.7, 60.7, 60.8, 60.8, 60.9, 60.9, 60.9, 60.9, 60.9, 60.9, 60.9, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.0, 61.1, 61.1, 61.1, 61.1, 61.1, 61.6, 61.7, 61.8, 64.5, 65.3, 65.3, 65.4, 69.0, 100.3, 100.6, 101.0, 101.5, 101.9, 101.9, 102.7, 103.7, 106.4, 106.4, 106.5, 106.5, 106.5, 106.5 ] }, { "x": 2.0, "output_throughput": { "median": 30.75, "min": 30.74, "max": 30.75, "n": 3 }, "median_ttft_ms": { "median": 140.87, "min": 140.0, "max": 140.89, "n": 3 }, "p99_ttft_ms": { "median": 148.64, "min": 148.08, "max": 153.66, "n": 3 }, "median_tpot_ms": { "median": 64.45, "min": 64.42, "max": 64.48, "n": 3 }, "p99_tpot_ms": { "median": 64.56, "min": 64.54, "max": 64.63, "n": 3 }, "p99_itl_ms": { "median": 65.1, "min": 65.05, "max": 65.12, "n": 3 }, "median_itl_ms": { "median": 64.48, "min": 64.47, "max": 64.49, "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": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 89.7, 89.7, 91.3, 96.0, 129.2, 129.3, 129.3, 129.3, 129.6, 130.3, 130.7, 135.1, 135.5, 135.5, 135.6, 135.7, 135.8, 135.8, 136.0, 136.2, 136.9, 139.9, 140.1, 140.4, 140.4, 140.4, 140.7, 140.7, 140.9, 140.9, 140.9, 141.0, 141.0, 141.1, 141.1, 141.1, 141.2, 141.4, 141.4, 141.8, 141.9, 141.9, 143.0, 143.2, 143.6, 145.1, 146.8, 147.4, 148.2, 149.6, 155.7 ], "tpot_quantiles_ms": [ 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, 64.4, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 64.5, 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 ] }, { "x": 4.0, "output_throughput": { "median": 60.65, "min": 60.65, "max": 60.68, "n": 3 }, "median_ttft_ms": { "median": 142.57, "min": 142.2, "max": 143.38, "n": 3 }, "p99_ttft_ms": { "median": 156.44, "min": 155.33, "max": 157.13, "n": 3 }, "median_tpot_ms": { "median": 65.3, "min": 65.28, "max": 65.31, "n": 3 }, "p99_tpot_ms": { "median": 65.36, "min": 65.36, "max": 65.38, "n": 3 }, "p99_itl_ms": { "median": 66.07, "min": 65.83, "max": 66.14, "n": 3 }, "median_itl_ms": { "median": 65.31, "min": 65.28, "max": 65.31, "n": 3 }, "goodput_rps": { "median": 0.47, "min": 0.47, "max": 0.47, "n": 3 }, "request_throughput": { "median": 0.47, "min": 0.47, "max": 0.47, "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": [ 94.8, 94.8, 95.2, 98.0, 133.4, 139.5, 140.3, 140.7, 140.7, 140.8, 140.9, 141.2, 141.3, 141.3, 141.4, 141.6, 141.8, 141.8, 141.9, 141.9, 141.9, 142.0, 142.0, 142.1, 142.4, 142.5, 142.5, 142.6, 143.0, 143.1, 143.1, 143.2, 143.7, 143.7, 143.7, 144.0, 144.6, 144.8, 145.1, 148.3, 149.2, 149.6, 154.5, 155.0, 155.4, 155.6, 156.1, 156.3, 156.5, 156.8, 157.2 ], "tpot_quantiles_ms": [ 65.2, 65.2, 65.2, 65.2, 65.2, 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, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.3, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4, 65.4 ] }, { "x": 8.0, "output_throughput": { "median": 111.22, "min": 111.16, "max": 111.36, "n": 3 }, "median_ttft_ms": { "median": 205.93, "min": 202.95, "max": 215.19, "n": 3 }, "p99_ttft_ms": { "median": 1261.21, "min": 1229.24, "max": 1293.53, "n": 3 }, "median_tpot_ms": { "median": 66.98, "min": 66.96, "max": 66.99, "n": 3 }, "p99_tpot_ms": { "median": 74.23, "min": 74.22, "max": 74.66, "n": 3 }, "p99_itl_ms": { "median": 68.19, "min": 68.06, "max": 68.27, "n": 3 }, "median_itl_ms": { "median": 67.0, "min": 66.99, "max": 67.02, "n": 3 }, "goodput_rps": { "median": 0.65, "min": 0.65, "max": 0.71, "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": 1515.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 102.7, 103.2, 143.0, 143.6, 144.2, 144.5, 145.0, 145.1, 145.6, 146.3, 146.8, 147.1, 147.6, 148.7, 149.8, 163.6, 164.1, 164.8, 165.2, 165.8, 166.4, 166.4, 166.8, 167.1, 167.5, 168.2, 244.7, 295.4, 304.8, 670.5, 686.0, 686.4, 746.5, 748.2, 755.1, 756.7, 777.9, 811.1, 812.4, 1167.3, 1168.4, 1179.4, 1180.6, 1201.9, 1202.8, 1228.1, 1229.4, 1259.6, 1260.7, 1292.7, 1293.7 ], "tpot_quantiles_ms": [ 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 66.9, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.0, 67.2, 69.8, 69.8, 70.2, 70.2, 70.3, 70.3, 71.0, 71.0, 71.0, 71.1, 71.1, 73.9, 74.3, 75.0 ] }, { "x": 16.0, "output_throughput": { "median": 198.37, "min": 198.24, "max": 198.38, "n": 3 }, "median_ttft_ms": { "median": 278.06, "min": 266.24, "max": 280.88, "n": 3 }, "p99_ttft_ms": { "median": 2448.76, "min": 2399.23, "max": 2457.84, "n": 3 }, "median_tpot_ms": { "median": 71.12, "min": 71.06, "max": 71.22, "n": 3 }, "p99_tpot_ms": { "median": 86.47, "min": 86.46, "max": 86.58, "n": 3 }, "p99_itl_ms": { "median": 79.67, "min": 79.58, "max": 85.27, "n": 3 }, "median_itl_ms": { "median": 71.05, "min": 71.0, "max": 71.07, "n": 3 }, "goodput_rps": { "median": 0.99, "min": 0.99, "max": 1.02, "n": 3 }, "request_throughput": { "median": 1.55, "min": 1.55, "max": 1.55, "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": [ 114.5, 156.3, 158.0, 160.6, 164.5, 171.8, 173.1, 175.1, 176.4, 177.7, 179.8, 183.7, 199.4, 200.7, 201.6, 202.1, 202.9, 203.2, 204.0, 205.4, 219.3, 220.4, 221.5, 222.5, 240.4, 253.4, 342.6, 812.2, 814.1, 820.3, 846.3, 849.8, 864.9, 1262.7, 1263.5, 1330.6, 1387.7, 1391.4, 1637.8, 1640.0, 1653.1, 1654.6, 1668.7, 1740.6, 1965.5, 1967.5, 1969.4, 2245.1, 2399.8, 2448.4, 2458.0 ], "tpot_quantiles_ms": [ 70.7, 70.7, 70.8, 70.8, 70.8, 70.8, 70.8, 70.8, 70.9, 70.9, 70.9, 70.9, 70.9, 70.9, 71.0, 71.0, 71.0, 71.0, 71.1, 71.1, 71.1, 71.1, 71.1, 71.1, 71.1, 71.2, 71.2, 71.2, 71.3, 75.3, 75.3, 75.4, 75.4, 75.4, 75.4, 76.1, 76.1, 76.2, 78.3, 78.4, 78.4, 79.4, 79.5, 82.3, 82.3, 82.3, 82.4, 82.6, 82.7, 86.4, 87.2 ] }, { "x": 32.0, "output_throughput": { "median": 273.22, "min": 272.36, "max": 273.34, "n": 3 }, "median_ttft_ms": { "median": 1767.61, "min": 1711.08, "max": 1793.45, "n": 3 }, "p99_ttft_ms": { "median": 4655.34, "min": 4650.57, "max": 4694.45, "n": 3 }, "median_tpot_ms": { "median": 103.11, "min": 102.88, "max": 103.35, "n": 3 }, "p99_tpot_ms": { "median": 114.32, "min": 114.28, "max": 114.9, "n": 3 }, "p99_itl_ms": { "median": 600.08, "min": 595.26, "max": 600.88, "n": 3 }, "median_itl_ms": { "median": 82.7, "min": 82.69, "max": 82.71, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.13, "min": 2.13, "max": 2.14, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1312.5, "min": 1305.0, "max": 1320.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 143.8, 385.6, 655.2, 788.5, 801.8, 806.1, 808.4, 1204.8, 1225.1, 1240.2, 1292.1, 1325.8, 1476.4, 1503.3, 1560.3, 1596.4, 1602.3, 1668.9, 1673.0, 1696.1, 1706.1, 1709.8, 1715.1, 1751.7, 1758.9, 1766.1, 1770.7, 1773.1, 1786.3, 1791.6, 1795.2, 1799.7, 1800.6, 1804.6, 1807.7, 1813.2, 1882.6, 2075.3, 2208.9, 2284.2, 2353.1, 2372.4, 2392.4, 2397.3, 2947.5, 2973.0, 3561.2, 4117.5, 4155.7, 4651.0, 4694.9 ], "tpot_quantiles_ms": [ 82.2, 82.3, 84.5, 88.7, 88.7, 92.9, 96.4, 97.0, 101.1, 101.8, 101.9, 102.0, 102.5, 102.5, 102.6, 102.7, 102.7, 102.7, 102.7, 102.8, 102.9, 102.9, 103.0, 103.0, 103.0, 103.1, 103.2, 103.3, 103.4, 103.5, 103.6, 103.9, 104.2, 104.3, 104.6, 104.9, 105.0, 105.2, 105.4, 105.8, 106.0, 107.8, 107.9, 108.5, 109.4, 109.8, 111.1, 111.8, 111.8, 113.7, 115.4 ] }, { "x": 64.0, "output_throughput": { "median": 242.49, "min": 241.32, "max": 245.56, "n": 3 }, "median_ttft_ms": { "median": 10940.97, "min": 10335.52, "max": 11433.22, "n": 3 }, "p99_ttft_ms": { "median": 26120.43, "min": 24482.1, "max": 27103.54, "n": 3 }, "median_tpot_ms": { "median": 131.42, "min": 128.48, "max": 133.76, "n": 3 }, "p99_tpot_ms": { "median": 259.5, "min": 258.25, "max": 263.6, "n": 3 }, "p99_itl_ms": { "median": 1325.5, "min": 1270.12, "max": 1427.29, "n": 3 }, "median_itl_ms": { "median": 85.97, "min": 85.91, "max": 85.97, "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": 1020.0, "min": 1005.0, "max": 1020.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.459 }, "ttft_quantiles_ms": [ 210.4, 1288.4, 1856.5, 2458.2, 3594.8, 4213.4, 4803.0, 4912.8, 4936.6, 5006.8, 5372.5, 5407.8, 5418.4, 5511.1, 5536.7, 5575.0, 5769.4, 5935.3, 6033.4, 6588.3, 7525.8, 8550.6, 8973.8, 9646.0, 10179.1, 10939.0, 11455.9, 11970.6, 12504.9, 13792.4, 15977.8, 16217.0, 16527.4, 16647.5, 16728.2, 16822.6, 16867.6, 17154.4, 18390.7, 18650.9, 19139.9, 19500.9, 19839.8, 20338.9, 21035.6, 22832.0, 23128.8, 23650.6, 24081.1, 24979.8, 27204.3 ], "tpot_quantiles_ms": [ 72.9, 79.9, 120.6, 120.9, 121.2, 121.6, 121.9, 122.0, 122.1, 122.4, 122.5, 122.7, 122.8, 122.8, 122.9, 123.1, 123.2, 123.3, 123.6, 123.8, 123.8, 124.4, 125.6, 127.3, 128.5, 131.0, 135.1, 139.7, 142.5, 145.2, 148.3, 148.9, 153.3, 158.6, 161.5, 162.8, 165.5, 169.0, 169.7, 172.3, 174.6, 175.6, 177.7, 179.7, 183.3, 186.9, 190.2, 197.3, 207.0, 241.6, 308.9 ] } ] }, "vllm-async": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 13.25, "min": 13.12, "max": 13.56, "n": 3 }, "median_ttft_ms": { "median": 191.25, "min": 189.23, "max": 194.35, "n": 3 }, "p99_ttft_ms": { "median": 695.16, "min": 684.3, "max": 705.37, "n": 3 }, "median_tpot_ms": { "median": 60.37, "min": 60.36, "max": 60.37, "n": 3 }, "p99_tpot_ms": { "median": 104.16, "min": 104.01, "max": 104.17, "n": 3 }, "p99_itl_ms": { "median": 104.72, "min": 104.46, "max": 104.85, "n": 3 }, "median_itl_ms": { "median": 60.47, "min": 60.47, "max": 60.5, "n": 3 }, "goodput_rps": { "median": 0.08, "min": 0.07, "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": 1267.5, "min": 1260.0, "max": 1282.5, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.486 }, "ttft_quantiles_ms": [ 76.6, 76.6, 77.1, 78.1, 167.2, 170.2, 171.9, 174.7, 178.7, 180.7, 181.6, 182.4, 183.8, 183.8, 183.9, 184.9, 185.2, 185.2, 185.8, 186.2, 186.2, 187.8, 188.1, 188.2, 190.2, 190.6, 190.9, 192.3, 192.9, 197.8, 211.1, 217.2, 344.3, 386.1, 386.1, 451.5, 462.6, 500.3, 547.9, 552.5, 555.8, 565.5, 566.7, 618.1, 628.1, 632.9, 640.2, 704.8, 704.9, 705.1, 705.4 ], "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, 60.4, 60.4, 60.4, 60.4, 60.4, 60.4, 60.6, 60.6, 60.8, 63.5, 64.2, 64.2, 76.9, 77.6, 78.6, 99.7, 99.8, 101.3, 103.2, 104.1, 104.1, 104.1, 104.2, 104.2, 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": 131.62, "min": 131.61, "max": 132.02, "n": 3 }, "p99_ttft_ms": { "median": 147.06, "min": 142.56, "max": 147.82, "n": 3 }, "median_tpot_ms": { "median": 63.43, "min": 63.43, "max": 63.44, "n": 3 }, "p99_tpot_ms": { "median": 63.55, "min": 63.54, "max": 63.56, "n": 3 }, "p99_itl_ms": { "median": 64.3, "min": 64.15, "max": 64.38, "n": 3 }, "median_itl_ms": { "median": 63.42, "min": 63.41, "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": 1507.5, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 79.0, 79.0, 79.2, 79.9, 128.2, 128.4, 128.6, 128.6, 128.6, 128.6, 128.7, 128.8, 128.8, 128.8, 128.8, 128.9, 128.9, 128.9, 129.0, 129.0, 129.0, 129.2, 129.3, 129.4, 129.7, 129.7, 133.5, 134.2, 134.3, 134.3, 134.5, 134.5, 134.6, 134.6, 134.6, 139.2, 139.5, 139.6, 139.7, 139.7, 139.8, 139.9, 139.9, 140.0, 140.1, 140.3, 140.5, 140.6, 142.9, 148.2, 149.2 ], "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, 63.4, 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, 63.6, 63.6 ] }, { "x": 4.0, "output_throughput": { "median": 61.79, "min": 61.78, "max": 61.81, "n": 3 }, "median_ttft_ms": { "median": 140.66, "min": 140.07, "max": 141.08, "n": 3 }, "p99_ttft_ms": { "median": 154.49, "min": 152.79, "max": 156.15, "n": 3 }, "median_tpot_ms": { "median": 64.1, "min": 64.09, "max": 64.11, "n": 3 }, "p99_tpot_ms": { "median": 64.2, "min": 64.19, "max": 64.21, "n": 3 }, "p99_itl_ms": { "median": 64.93, "min": 64.82, "max": 64.99, "n": 3 }, "median_itl_ms": { "median": 64.11, "min": 64.09, "max": 64.12, "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": 1545.0, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 82.5, 82.5, 83.6, 85.4, 131.6, 131.6, 131.9, 132.2, 132.2, 132.4, 132.6, 132.7, 134.0, 139.5, 139.6, 139.6, 139.9, 139.9, 139.9, 140.1, 140.1, 140.2, 140.2, 140.4, 140.4, 140.5, 140.6, 140.7, 140.8, 140.8, 140.9, 141.0, 141.0, 141.1, 141.1, 141.2, 141.4, 141.4, 141.5, 141.8, 142.2, 143.1, 151.9, 152.5, 152.8, 153.6, 154.2, 154.5, 155.3, 155.8, 156.2 ], "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, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2 ] }, { "x": 8.0, "output_throughput": { "median": 112.68, "min": 112.58, "max": 112.84, "n": 3 }, "median_ttft_ms": { "median": 261.27, "min": 250.59, "max": 261.68, "n": 3 }, "p99_ttft_ms": { "median": 1328.81, "min": 1311.19, "max": 1336.29, "n": 3 }, "median_tpot_ms": { "median": 65.75, "min": 65.69, "max": 65.85, "n": 3 }, "p99_tpot_ms": { "median": 73.53, "min": 73.29, "max": 73.55, "n": 3 }, "p99_itl_ms": { "median": 67.3, "min": 67.08, "max": 67.94, "n": 3 }, "median_itl_ms": { "median": 65.69, "min": 65.65, "max": 65.74, "n": 3 }, "goodput_rps": { "median": 0.66, "min": 0.61, "max": 0.69, "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": 1470.0, "max": 1500.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.1, 87.0, 134.6, 134.9, 138.1, 138.7, 139.7, 162.2, 162.8, 163.1, 163.6, 163.7, 164.0, 164.4, 164.9, 165.3, 165.9, 166.5, 211.3, 212.0, 212.5, 212.6, 212.8, 213.2, 213.8, 214.3, 308.2, 312.6, 339.2, 571.6, 573.2, 579.9, 589.5, 711.3, 712.5, 766.0, 974.5, 975.5, 1086.8, 1087.4, 1089.1, 1097.4, 1098.9, 1114.7, 1151.1, 1152.2, 1181.1, 1289.2, 1290.2, 1311.3, 1357.0 ], "tpot_quantiles_ms": [ 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, 65.6, 65.6, 65.6, 65.6, 65.6, 65.6, 65.7, 65.8, 65.9, 66.6, 66.9, 66.9, 66.9, 66.9, 67.0, 67.0, 67.0, 67.6, 67.6, 67.9, 69.9, 69.9, 70.5, 70.5, 70.9, 70.9, 71.2, 71.7, 73.0, 73.5, 73.8 ] }, { "x": 16.0, "output_throughput": { "median": 201.37, "min": 200.77, "max": 201.51, "n": 3 }, "median_ttft_ms": { "median": 284.17, "min": 282.56, "max": 289.25, "n": 3 }, "p99_ttft_ms": { "median": 2472.23, "min": 2465.48, "max": 2550.45, "n": 3 }, "median_tpot_ms": { "median": 69.49, "min": 69.46, "max": 69.61, "n": 3 }, "p99_tpot_ms": { "median": 85.44, "min": 85.42, "max": 85.98, "n": 3 }, "p99_itl_ms": { "median": 84.86, "min": 78.82, "max": 86.56, "n": 3 }, "median_itl_ms": { "median": 69.5, "min": 69.41, "max": 69.52, "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": 1410.0, "max": 1455.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 98.9, 173.9, 175.2, 176.5, 177.7, 190.9, 191.9, 193.0, 194.4, 202.8, 204.8, 216.4, 217.5, 220.2, 221.8, 223.6, 225.6, 227.3, 229.6, 238.1, 243.1, 251.7, 252.6, 253.0, 253.9, 266.4, 346.4, 825.1, 838.6, 863.1, 879.7, 880.8, 886.5, 1305.4, 1309.4, 1372.0, 1412.8, 1473.8, 1690.7, 1692.5, 1695.4, 1697.7, 1700.2, 1775.9, 1959.3, 1998.7, 2026.2, 2299.9, 2464.6, 2506.1, 2550.7 ], "tpot_quantiles_ms": [ 69.2, 69.2, 69.3, 69.3, 69.3, 69.3, 69.3, 69.3, 69.3, 69.3, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.4, 69.5, 69.5, 69.5, 69.6, 69.6, 69.6, 69.9, 73.7, 73.8, 74.2, 74.2, 74.2, 74.2, 74.9, 75.0, 75.0, 77.2, 77.2, 77.5, 77.9, 78.4, 81.2, 81.3, 81.3, 81.3, 81.7, 81.7, 85.5, 86.1 ] }, { "x": 32.0, "output_throughput": { "median": 275.3, "min": 273.78, "max": 275.31, "n": 3 }, "median_ttft_ms": { "median": 1806.54, "min": 1769.74, "max": 1854.26, "n": 3 }, "p99_ttft_ms": { "median": 4815.8, "min": 4732.89, "max": 4874.22, "n": 3 }, "median_tpot_ms": { "median": 102.1, "min": 101.77, "max": 102.1, "n": 3 }, "p99_tpot_ms": { "median": 113.64, "min": 113.56, "max": 114.09, "n": 3 }, "p99_itl_ms": { "median": 613.22, "min": 611.14, "max": 619.43, "n": 3 }, "median_itl_ms": { "median": 80.57, "min": 80.57, "max": 80.58, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.15, "min": 2.14, "max": 2.15, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1275.0, "min": 1275.0, "max": 1290.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 118.8, 392.6, 673.0, 813.2, 827.9, 837.1, 840.3, 1259.7, 1273.9, 1276.9, 1330.0, 1379.4, 1516.4, 1571.1, 1621.5, 1663.1, 1671.0, 1724.4, 1730.2, 1735.0, 1739.3, 1768.9, 1775.8, 1792.8, 1796.0, 1806.0, 1823.1, 1829.1, 1831.6, 1838.0, 1843.9, 1857.5, 1861.4, 1864.2, 1869.5, 1876.9, 1951.5, 2147.7, 2286.2, 2343.6, 2400.5, 2438.2, 2466.8, 2474.6, 3026.9, 3074.9, 3663.8, 4219.0, 4282.5, 4733.4, 4874.7 ], "tpot_quantiles_ms": [ 80.2, 80.3, 82.7, 87.0, 87.1, 91.4, 95.1, 95.7, 100.0, 100.5, 100.6, 100.7, 100.8, 101.1, 101.2, 101.2, 101.3, 101.4, 101.5, 101.6, 101.6, 101.6, 101.8, 101.9, 102.0, 102.1, 102.1, 102.1, 102.3, 102.4, 102.4, 102.6, 102.9, 103.2, 103.4, 103.6, 103.9, 104.1, 104.4, 104.5, 105.0, 106.4, 106.9, 107.7, 108.0, 108.6, 110.3, 110.6, 111.2, 113.0, 114.7 ] }, { "x": 64.0, "output_throughput": { "median": 241.97, "min": 241.2, "max": 253.32, "n": 3 }, "median_ttft_ms": { "median": 11087.23, "min": 9801.72, "max": 11174.74, "n": 3 }, "p99_ttft_ms": { "median": 25950.32, "min": 23667.81, "max": 27243.21, "n": 3 }, "median_tpot_ms": { "median": 128.17, "min": 126.45, "max": 129.68, "n": 3 }, "p99_tpot_ms": { "median": 259.59, "min": 241.76, "max": 267.98, "n": 3 }, "p99_itl_ms": { "median": 1127.44, "min": 1115.96, "max": 1423.48, "n": 3 }, "median_itl_ms": { "median": 83.91, "min": 83.62, "max": 84.16, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 1.89, "min": 1.88, "max": 1.98, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 997.5, "min": 990.0, "max": 1005.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.44 }, "ttft_quantiles_ms": [ 170.6, 1332.9, 1918.5, 2531.4, 3523.4, 4341.0, 4946.7, 4967.7, 4989.4, 5068.1, 5319.3, 5484.4, 5510.1, 5569.5, 5589.5, 5610.0, 5680.0, 5895.8, 6191.0, 6777.0, 7030.1, 8277.8, 8539.0, 9337.3, 9807.3, 10473.0, 11095.7, 11489.5, 12313.7, 13381.6, 15731.5, 16026.0, 16334.7, 16506.3, 16602.7, 16652.2, 16678.0, 16955.0, 17411.5, 18077.5, 18626.6, 19053.2, 19554.9, 20261.5, 20870.1, 22382.1, 22997.6, 23643.9, 24690.0, 25397.6, 27265.8 ], "tpot_quantiles_ms": [ 69.2, 78.6, 118.8, 119.2, 120.1, 120.6, 120.8, 120.9, 121.0, 121.2, 121.3, 121.5, 121.7, 122.1, 122.2, 122.3, 122.4, 122.6, 122.6, 122.8, 123.6, 123.7, 124.0, 125.3, 126.3, 127.2, 129.5, 131.1, 135.0, 138.2, 144.1, 145.6, 154.3, 158.7, 160.9, 163.2, 166.4, 168.2, 169.1, 171.4, 172.4, 174.9, 177.8, 181.4, 184.8, 187.7, 191.3, 195.8, 208.1, 237.9, 312.9 ] } ] }, "vllm-seqs-16": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 13.35, "min": 13.11, "max": 13.63, "n": 3 }, "median_ttft_ms": { "median": 190.07, "min": 187.02, "max": 195.64, "n": 3 }, "p99_ttft_ms": { "median": 688.21, "min": 613.66, "max": 696.43, "n": 3 }, "median_tpot_ms": { "median": 60.35, "min": 60.35, "max": 60.36, "n": 3 }, "p99_tpot_ms": { "median": 104.18, "min": 104.16, "max": 104.2, "n": 3 }, "p99_itl_ms": { "median": 104.57, "min": 104.57, "max": 104.74, "n": 3 }, "median_itl_ms": { "median": 60.45, "min": 60.45, "max": 60.46, "n": 3 }, "goodput_rps": { "median": 0.08, "min": 0.07, "max": 0.08, "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": 1275.0, "max": 1290.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.48 }, "ttft_quantiles_ms": [ 76.3, 76.3, 76.4, 77.1, 172.5, 173.1, 174.8, 177.8, 181.0, 183.8, 183.8, 184.1, 184.2, 184.2, 184.9, 184.9, 185.2, 185.2, 186.1, 186.2, 186.5, 186.5, 187.6, 187.8, 188.8, 189.6, 189.9, 190.3, 193.7, 196.2, 197.6, 211.6, 227.4, 359.6, 359.6, 420.1, 425.7, 427.8, 475.3, 475.9, 536.6, 583.9, 593.1, 611.6, 611.8, 614.0, 642.0, 643.5, 649.3, 704.7, 705.0 ], "tpot_quantiles_ms": [ 60.2, 60.2, 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, 60.4, 60.4, 60.4, 60.6, 60.7, 62.2, 63.1, 63.4, 63.4, 64.0, 73.4, 73.6, 104.0, 104.1, 104.1, 104.1, 104.1, 104.1, 104.2, 104.2, 104.2, 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": 132.21, "min": 131.63, "max": 132.57, "n": 3 }, "p99_ttft_ms": { "median": 147.54, "min": 142.16, "max": 149.38, "n": 3 }, "median_tpot_ms": { "median": 63.43, "min": 63.42, "max": 63.44, "n": 3 }, "p99_tpot_ms": { "median": 63.55, "min": 63.52, "max": 63.56, "n": 3 }, "p99_itl_ms": { "median": 64.55, "min": 64.25, "max": 64.59, "n": 3 }, "median_itl_ms": { "median": 63.42, "min": 63.41, "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": 1500.0, "n": 3 }, "max_temp_c": 87.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 79.5, 79.5, 79.5, 81.9, 128.0, 128.5, 128.6, 128.8, 128.8, 128.8, 128.9, 129.0, 129.2, 129.2, 129.2, 129.2, 129.2, 129.2, 129.3, 129.3, 129.3, 129.5, 129.7, 129.8, 130.2, 130.3, 133.9, 134.6, 134.6, 134.8, 134.9, 135.3, 135.9, 136.4, 136.4, 139.4, 139.4, 139.6, 139.6, 139.7, 139.7, 139.8, 139.9, 140.0, 140.1, 140.3, 140.6, 140.7, 142.5, 148.7, 150.9 ], "tpot_quantiles_ms": [ 63.3, 63.3, 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, 63.4, 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, 63.6, 63.6 ] }, { "x": 4.0, "output_throughput": { "median": 61.58, "min": 61.34, "max": 61.58, "n": 3 }, "median_ttft_ms": { "median": 142.18, "min": 141.53, "max": 193.84, "n": 3 }, "p99_ttft_ms": { "median": 156.44, "min": 153.79, "max": 209.61, "n": 3 }, "median_tpot_ms": { "median": 64.32, "min": 64.24, "max": 64.33, "n": 3 }, "p99_tpot_ms": { "median": 64.41, "min": 64.4, "max": 64.41, "n": 3 }, "p99_itl_ms": { "median": 65.4, "min": 65.3, "max": 65.61, "n": 3 }, "median_itl_ms": { "median": 64.31, "min": 64.21, "max": 64.32, "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": 1537.5, "n": 3 }, "max_temp_c": 82.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 81.3, 81.3, 82.8, 84.0, 132.3, 132.6, 133.3, 133.4, 133.9, 134.1, 134.8, 138.3, 139.8, 140.6, 141.1, 141.1, 141.1, 141.1, 141.5, 141.6, 141.8, 142.0, 142.0, 142.4, 142.5, 142.8, 142.9, 143.1, 143.2, 143.6, 143.7, 152.9, 153.5, 153.8, 153.8, 153.9, 154.4, 154.8, 155.6, 156.2, 156.5, 190.7, 191.1, 196.6, 197.2, 204.6, 205.1, 205.4, 205.9, 209.1, 209.7 ], "tpot_quantiles_ms": [ 64.1, 64.1, 64.1, 64.1, 64.1, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.2, 64.3, 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, 64.3, 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 ] }, { "x": 8.0, "output_throughput": { "median": 112.47, "min": 112.41, "max": 112.58, "n": 3 }, "median_ttft_ms": { "median": 259.54, "min": 258.41, "max": 259.69, "n": 3 }, "p99_ttft_ms": { "median": 1322.42, "min": 1315.26, "max": 1367.78, "n": 3 }, "median_tpot_ms": { "median": 65.91, "min": 65.89, "max": 66.01, "n": 3 }, "p99_tpot_ms": { "median": 73.48, "min": 73.42, "max": 73.52, "n": 3 }, "p99_itl_ms": { "median": 67.25, "min": 67.12, "max": 67.41, "n": 3 }, "median_itl_ms": { "median": 65.89, "min": 65.86, "max": 65.93, "n": 3 }, "goodput_rps": { "median": 0.66, "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": 1515.0, "n": 3 }, "max_temp_c": 83.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 86.6, 86.7, 135.2, 135.5, 138.3, 140.1, 142.1, 143.2, 144.0, 162.4, 162.8, 162.8, 163.1, 163.4, 163.7, 163.8, 164.1, 164.3, 164.5, 164.8, 211.3, 211.4, 212.0, 212.7, 213.8, 215.3, 304.5, 306.1, 334.8, 567.7, 569.9, 703.0, 704.5, 713.8, 841.0, 842.7, 865.6, 1082.5, 1083.5, 1091.2, 1092.1, 1177.7, 1241.3, 1261.4, 1262.5, 1263.8, 1272.0, 1306.6, 1307.8, 1349.4, 1367.9 ], "tpot_quantiles_ms": [ 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, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 65.9, 66.0, 66.1, 66.8, 66.8, 67.1, 67.6, 69.3, 69.3, 69.4, 69.4, 70.0, 70.2, 70.2, 70.9, 71.7, 73.1, 73.5, 73.8 ] }, { "x": 16.0, "output_throughput": { "median": 200.87, "min": 200.61, "max": 201.13, "n": 3 }, "median_ttft_ms": { "median": 287.94, "min": 285.62, "max": 295.66, "n": 3 }, "p99_ttft_ms": { "median": 2481.61, "min": 2459.89, "max": 2547.79, "n": 3 }, "median_tpot_ms": { "median": 69.66, "min": 69.63, "max": 69.82, "n": 3 }, "p99_tpot_ms": { "median": 85.56, "min": 85.54, "max": 85.92, "n": 3 }, "p99_itl_ms": { "median": 80.38, "min": 78.77, "max": 85.56, "n": 3 }, "median_itl_ms": { "median": 69.56, "min": 69.44, "max": 69.61, "n": 3 }, "goodput_rps": { "median": 1.01, "min": 1.01, "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": 1440.0, "min": 1425.0, "max": 1455.0, "n": 3 }, "max_temp_c": 84.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 101.2, 185.3, 186.4, 187.9, 190.8, 192.1, 192.8, 204.6, 205.8, 207.0, 208.3, 218.7, 219.4, 220.4, 221.4, 222.6, 223.5, 224.3, 226.5, 230.1, 243.8, 245.0, 263.5, 264.4, 266.8, 294.7, 351.6, 836.3, 839.2, 842.6, 879.7, 908.8, 911.6, 1306.6, 1337.5, 1371.8, 1428.8, 1438.1, 1691.1, 1694.0, 1695.8, 1717.7, 1746.4, 1771.9, 1973.0, 2015.5, 2030.6, 2371.2, 2460.2, 2481.9, 2548.1 ], "tpot_quantiles_ms": [ 69.2, 69.3, 69.3, 69.3, 69.3, 69.4, 69.4, 69.4, 69.4, 69.5, 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, 69.9, 69.9, 73.6, 73.9, 74.1, 74.1, 74.2, 74.2, 74.9, 75.1, 75.1, 77.0, 77.2, 77.3, 77.8, 78.3, 81.1, 81.2, 81.3, 81.5, 81.7, 81.8, 85.5, 86.4 ] }, { "x": 32.0, "output_throughput": { "median": 171.78, "min": 171.27, "max": 172.45, "n": 3 }, "median_ttft_ms": { "median": 12330.02, "min": 12300.65, "max": 12349.65, "n": 3 }, "p99_ttft_ms": { "median": 16060.48, "min": 15956.19, "max": 16402.26, "n": 3 }, "median_tpot_ms": { "median": 81.14, "min": 81.03, "max": 81.53, "n": 3 }, "p99_tpot_ms": { "median": 103.79, "min": 102.41, "max": 105.16, "n": 3 }, "p99_itl_ms": { "median": 594.51, "min": 587.4, "max": 602.86, "n": 3 }, "median_itl_ms": { "median": 69.79, "min": 69.78, "max": 69.82, "n": 3 }, "goodput_rps": { "median": 0.04, "min": 0.04, "max": 0.05, "n": 3 }, "request_throughput": { "median": 1.34, "min": 1.34, "max": 1.35, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 1365.0, "min": 1335.0, "max": 1365.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.22 }, "ttft_quantiles_ms": [ 127.1, 676.6, 1266.8, 1275.4, 1845.9, 2329.0, 2361.8, 11463.6, 11589.6, 11692.3, 11819.2, 11923.6, 11976.7, 12010.9, 12047.1, 12130.7, 12184.6, 12245.0, 12251.1, 12282.4, 12286.0, 12292.5, 12301.2, 12305.2, 12310.6, 12313.0, 12334.3, 12351.2, 12369.2, 12401.9, 12434.0, 12492.9, 12537.3, 12680.2, 12794.1, 12830.4, 12871.0, 12905.1, 13160.8, 13250.5, 13422.0, 13473.1, 13611.7, 13999.1, 14456.3, 14750.0, 15387.0, 15791.3, 15870.3, 15992.0, 16404.2 ], "tpot_quantiles_ms": [ 70.5, 73.0, 77.5, 77.8, 77.9, 78.1, 78.2, 78.3, 78.4, 78.7, 78.8, 78.9, 79.0, 79.4, 79.7, 80.0, 80.0, 80.1, 80.1, 80.7, 80.7, 80.8, 80.9, 81.0, 81.3, 81.5, 81.5, 81.8, 82.1, 82.6, 82.6, 82.9, 83.1, 83.4, 84.1, 84.8, 85.3, 86.0, 86.7, 87.8, 88.4, 89.9, 93.0, 93.6, 94.2, 94.7, 95.1, 96.8, 98.3, 102.4, 113.6 ] }, { "x": 64.0, "output_throughput": { "median": 138.08, "min": 136.87, "max": 139.35, "n": 3 }, "median_ttft_ms": { "median": 39889.68, "min": 39484.5, "max": 40521.88, "n": 3 }, "p99_ttft_ms": { "median": 58190.26, "min": 56789.37, "max": 58350.91, "n": 3 }, "median_tpot_ms": { "median": 103.7, "min": 103.5, "max": 106.48, "n": 3 }, "p99_tpot_ms": { "median": 137.83, "min": 132.84, "max": 138.09, "n": 3 }, "p99_itl_ms": { "median": 1142.95, "min": 1086.08, "max": 1281.85, "n": 3 }, "median_itl_ms": { "median": 70.56, "min": 70.37, "max": 70.56, "n": 3 }, "goodput_rps": { "median": 0.02, "min": 0.02, "max": 0.02, "n": 3 }, "request_throughput": { "median": 1.08, "min": 1.07, "max": 1.09, "n": 3 }, "completed": [ 256, 256, 256 ], "gpu": { "median_sm_mhz": { "median": 630.0, "min": 487.5, "max": 915.0, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.656 }, "ttft_quantiles_ms": [ 166.6, 1305.9, 1908.6, 2430.1, 12350.6, 12929.5, 13578.8, 23016.1, 23955.1, 24700.9, 34031.7, 34304.0, 34506.8, 34584.1, 34780.6, 34872.0, 35015.8, 35121.2, 35388.4, 35586.4, 35842.6, 36223.6, 36833.2, 38173.3, 39243.7, 40133.0, 42081.7, 43587.6, 44552.5, 47252.1, 48305.4, 49506.4, 51345.5, 52171.1, 52637.7, 53358.2, 54229.2, 54282.7, 54512.9, 54911.2, 55164.2, 55349.2, 55594.9, 55791.4, 56278.0, 56621.8, 56972.3, 57499.1, 57620.7, 58130.3, 58539.4 ], "tpot_quantiles_ms": [ 77.1, 77.7, 78.1, 78.2, 78.7, 78.9, 79.5, 79.9, 80.1, 80.6, 80.8, 81.3, 81.5, 82.0, 82.5, 82.9, 84.0, 85.1, 86.3, 87.8, 91.5, 94.3, 96.5, 99.1, 101.6, 103.7, 107.3, 108.9, 109.4, 110.9, 111.7, 113.3, 114.4, 115.0, 116.1, 116.7, 117.4, 118.4, 119.0, 119.7, 120.7, 121.4, 122.5, 122.8, 123.0, 124.3, 125.4, 127.4, 129.0, 131.0, 142.4 ] } ] }, "vllm-fast-b": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 10.59, "min": 10.44, "max": 10.69, "n": 3 }, "median_ttft_ms": { "median": 704.63, "min": 702.94, "max": 704.94, "n": 3 }, "p99_ttft_ms": { "median": 706.74, "min": 706.13, "max": 710.82, "n": 3 }, "median_tpot_ms": { "median": 104.14, "min": 104.09, "max": 104.15, "n": 3 }, "p99_tpot_ms": { "median": 104.19, "min": 104.16, "max": 104.2, "n": 3 }, "p99_itl_ms": { "median": 104.72, "min": 104.53, "max": 104.83, "n": 3 }, "median_itl_ms": { "median": 104.06, "min": 104.04, "max": 104.08, "n": 3 }, "goodput_rps": { "median": 0.03, "min": 0.03, "max": 0.03, "n": 3 }, "request_throughput": { "median": 0.08, "min": 0.08, "max": 0.08, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 210.0, "min": 210.0, "max": 210.0, "n": 3 }, "max_temp_c": 93.0, "thermal_share": 0.787 }, "ttft_quantiles_ms": [ 75.5, 75.5, 76.8, 77.1, 176.5, 177.6, 184.5, 187.8, 188.4, 190.6, 194.1, 194.1, 217.0, 238.9, 410.5, 427.3, 483.9, 483.9, 491.2, 586.7, 616.6, 639.5, 698.1, 702.7, 703.1, 703.4, 703.8, 704.5, 704.6, 704.6, 704.8, 704.9, 704.9, 705.1, 705.1, 705.1, 705.4, 705.4, 705.7, 705.9, 706.0, 706.0, 706.1, 706.1, 706.6, 706.9, 707.0, 708.2, 708.3, 708.8, 711.2 ], "tpot_quantiles_ms": [ 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.4, 60.4, 60.4, 60.5, 62.3, 62.5, 72.5, 73.7, 86.1, 86.1, 86.1, 97.8, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2 ] }, { "x": 32.0, "output_throughput": { "median": 186.68, "min": 185.41, "max": 187.09, "n": 3 }, "median_ttft_ms": { "median": 4180.76, "min": 4131.39, "max": 4699.15, "n": 3 }, "p99_ttft_ms": { "median": 7319.11, "min": 7180.95, "max": 8018.23, "n": 3 }, "median_tpot_ms": { "median": 142.67, "min": 142.09, "max": 148.69, "n": 3 }, "p99_tpot_ms": { "median": 205.64, "min": 204.06, "max": 212.09, "n": 3 }, "p99_itl_ms": { "median": 1705.44, "min": 1424.28, "max": 1788.95, "n": 3 }, "median_itl_ms": { "median": 84.75, "min": 83.68, "max": 105.78, "n": 3 }, "goodput_rps": { "median": 0.22, "min": 0.22, "max": 0.22, "n": 3 }, "request_throughput": { "median": 1.46, "min": 1.45, "max": 1.46, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 450.0, "min": 435.0, "max": 562.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.694 }, "ttft_quantiles_ms": [ 118.8, 195.3, 203.7, 220.4, 390.0, 787.5, 793.9, 810.6, 837.1, 1156.6, 1414.9, 1448.1, 2030.6, 2037.5, 2214.3, 2294.6, 2535.4, 2655.9, 2689.3, 2839.0, 3220.2, 3806.4, 3933.5, 4074.6, 4129.7, 4180.9, 4199.5, 4339.3, 4528.3, 4837.1, 4951.6, 5106.6, 5172.2, 5253.0, 5356.1, 5368.0, 5401.6, 5464.9, 5483.4, 5519.6, 5699.0, 5800.2, 5912.2, 6041.6, 6125.1, 6176.8, 6514.1, 6799.7, 7044.2, 7316.8, 8154.4 ], "tpot_quantiles_ms": [ 90.2, 90.3, 92.0, 93.4, 93.5, 93.5, 94.6, 94.7, 95.8, 97.5, 97.7, 97.7, 98.7, 111.3, 114.4, 124.2, 128.0, 131.1, 132.5, 134.3, 139.0, 139.6, 141.2, 142.1, 142.4, 146.0, 148.2, 149.8, 151.5, 153.4, 154.3, 156.1, 159.9, 161.6, 164.4, 166.3, 166.8, 167.6, 168.3, 169.4, 171.0, 174.5, 177.4, 177.8, 181.4, 182.7, 190.1, 192.3, 199.7, 205.6, 224.7 ] } ] }, "vllm-O0": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 10.47, "min": 10.35, "max": 10.59, "n": 2 }, "median_ttft_ms": { "median": 707.14, "min": 707.04, "max": 707.25, "n": 2 }, "p99_ttft_ms": { "median": 712.44, "min": 711.42, "max": 713.46, "n": 2 }, "median_tpot_ms": { "median": 106.48, "min": 106.48, "max": 106.48, "n": 2 }, "p99_tpot_ms": { "median": 106.51, "min": 106.5, "max": 106.51, "n": 2 }, "p99_itl_ms": { "median": 107.05, "min": 106.98, "max": 107.13, "n": 2 }, "median_itl_ms": { "median": 106.32, "min": 106.31, "max": 106.34, "n": 2 }, "goodput_rps": { "median": 0.03, "min": 0.03, "max": 0.04, "n": 2 }, "request_throughput": { "median": 0.08, "min": 0.08, "max": 0.08, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 210.0, "min": 210.0, "max": 210.0, "n": 2 }, "max_temp_c": 93.0, "thermal_share": 0.783 }, "ttft_quantiles_ms": [ 89.7, 89.7, 95.9, 95.9, 184.0, 184.8, 184.8, 186.2, 186.2, 186.9, 193.9, 193.9, 210.9, 303.5, 303.5, 353.4, 353.4, 506.3, 510.3, 510.3, 528.3, 653.5, 653.5, 706.4, 706.4, 707.1, 707.4, 707.4, 707.4, 707.4, 707.7, 707.7, 707.7, 708.4, 708.9, 708.9, 709.0, 709.0, 709.8, 710.0, 710.0, 710.3, 710.8, 710.8, 711.0, 711.0, 711.3, 711.5, 711.5, 712.3, 713.7 ], "tpot_quantiles_ms": [ 60.9, 60.9, 60.9, 60.9, 61.0, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.4, 68.3, 68.3, 89.8, 89.8, 90.5, 91.0, 91.0, 96.0, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5, 106.5 ] }, { "x": 32.0, "output_throughput": { "median": 191.53, "min": 190.95, "max": 192.11, "n": 2 }, "median_ttft_ms": { "median": 4051.14, "min": 4030.96, "max": 4071.31, "n": 2 }, "p99_ttft_ms": { "median": 7460.57, "min": 7447.38, "max": 7473.76, "n": 2 }, "median_tpot_ms": { "median": 142.04, "min": 141.74, "max": 142.35, "n": 2 }, "p99_tpot_ms": { "median": 201.42, "min": 199.59, "max": 203.24, "n": 2 }, "p99_itl_ms": { "median": 1459.42, "min": 1354.17, "max": 1564.68, "n": 2 }, "median_itl_ms": { "median": 85.61, "min": 85.22, "max": 85.99, "n": 2 }, "goodput_rps": { "median": 0.22, "min": 0.22, "max": 0.22, "n": 2 }, "request_throughput": { "median": 1.5, "min": 1.49, "max": 1.5, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 547.5, "min": 487.5, "max": 607.5, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.676 }, "ttft_quantiles_ms": [ 161.6, 699.1, 700.6, 702.2, 703.6, 705.6, 707.2, 708.7, 865.5, 1120.6, 1289.6, 1306.7, 1884.4, 2027.6, 2094.7, 2377.7, 2494.2, 2507.8, 2612.0, 2959.7, 3048.0, 3889.5, 3898.9, 4008.9, 4023.8, 4031.7, 4055.0, 4072.5, 4087.7, 4151.0, 4232.7, 4622.1, 4740.1, 4817.3, 4977.4, 5010.2, 5168.5, 5376.7, 5445.1, 5471.4, 5606.1, 5666.7, 5691.9, 5720.2, 5785.2, 5977.7, 6139.1, 6632.8, 6653.7, 7081.3, 7653.4 ], "tpot_quantiles_ms": [ 88.6, 89.2, 90.4, 91.6, 92.1, 95.8, 95.8, 95.8, 95.8, 96.3, 96.3, 96.3, 98.5, 109.1, 111.3, 120.8, 122.0, 126.6, 128.6, 131.7, 133.0, 133.6, 134.9, 138.1, 141.7, 142.3, 142.8, 144.9, 145.4, 148.0, 148.8, 150.5, 152.9, 159.3, 160.3, 161.5, 162.9, 163.6, 165.0, 165.8, 168.0, 171.1, 172.3, 176.0, 176.4, 177.1, 181.2, 187.3, 191.3, 199.6, 215.2 ] } ] }, "vllm-O3": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 10.63, "min": 10.54, "max": 10.72, "n": 2 }, "median_ttft_ms": { "median": 704.76, "min": 704.37, "max": 705.14, "n": 2 }, "p99_ttft_ms": { "median": 709.72, "min": 709.53, "max": 709.9, "n": 2 }, "median_tpot_ms": { "median": 104.15, "min": 104.15, "max": 104.16, "n": 2 }, "p99_tpot_ms": { "median": 104.19, "min": 104.18, "max": 104.2, "n": 2 }, "p99_itl_ms": { "median": 104.74, "min": 104.61, "max": 104.87, "n": 2 }, "median_itl_ms": { "median": 104.07, "min": 104.07, "max": 104.07, "n": 2 }, "goodput_rps": { "median": 0.03, "min": 0.03, "max": 0.03, "n": 2 }, "request_throughput": { "median": 0.08, "min": 0.08, "max": 0.08, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 210.0, "min": 210.0, "max": 210.0, "n": 2 }, "max_temp_c": 94.0, "thermal_share": 0.784 }, "ttft_quantiles_ms": [ 77.0, 77.0, 77.1, 77.1, 181.3, 189.3, 189.3, 190.4, 190.4, 191.3, 195.7, 195.7, 217.6, 313.2, 313.2, 466.9, 466.9, 472.2, 491.7, 491.7, 508.8, 702.9, 702.9, 703.3, 703.3, 704.2, 704.5, 704.5, 705.0, 705.0, 706.0, 706.3, 706.3, 706.3, 706.9, 706.9, 707.0, 707.0, 707.3, 707.5, 707.5, 707.5, 707.6, 707.6, 707.7, 707.7, 708.1, 709.4, 709.4, 709.9, 710.0 ], "tpot_quantiles_ms": [ 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.3, 60.4, 60.4, 60.4, 62.6, 71.2, 71.2, 82.4, 82.4, 87.1, 94.6, 94.6, 100.6, 104.1, 104.1, 104.1, 104.1, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2, 104.2 ] }, { "x": 32.0, "output_throughput": { "median": 189.86, "min": 188.5, "max": 191.21, "n": 2 }, "median_ttft_ms": { "median": 4086.24, "min": 3988.31, "max": 4184.17, "n": 2 }, "p99_ttft_ms": { "median": 6681.01, "min": 6214.24, "max": 7147.77, "n": 2 }, "median_tpot_ms": { "median": 143.85, "min": 139.59, "max": 148.11, "n": 2 }, "p99_tpot_ms": { "median": 201.71, "min": 201.48, "max": 201.95, "n": 2 }, "p99_itl_ms": { "median": 1601.94, "min": 1416.58, "max": 1787.29, "n": 2 }, "median_itl_ms": { "median": 94.23, "min": 83.87, "max": 104.58, "n": 2 }, "goodput_rps": { "median": 0.22, "min": 0.22, "max": 0.22, "n": 2 }, "request_throughput": { "median": 1.48, "min": 1.47, "max": 1.49, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 536.25, "min": 450.0, "max": 622.5, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.691 }, "ttft_quantiles_ms": [ 115.8, 207.1, 208.8, 232.2, 233.4, 235.4, 806.0, 807.5, 829.2, 1054.2, 1429.7, 1461.5, 1970.6, 2054.1, 2083.8, 2219.7, 2640.6, 2673.5, 2820.0, 2875.4, 2970.9, 3626.3, 3793.9, 3813.5, 3990.2, 4134.0, 4184.0, 4186.8, 4232.5, 4254.2, 4266.1, 4443.6, 4590.1, 4663.6, 4758.5, 4806.3, 4840.4, 5111.4, 5248.5, 5343.7, 5353.1, 5386.4, 5390.9, 5553.3, 5561.0, 5670.4, 5791.7, 6142.2, 6272.1, 6645.0, 7187.5 ], "tpot_quantiles_ms": [ 90.2, 91.2, 91.3, 93.0, 93.6, 94.4, 94.4, 96.8, 97.6, 97.7, 98.5, 98.5, 98.6, 113.8, 119.4, 121.8, 122.6, 127.1, 127.3, 132.9, 136.1, 136.6, 139.6, 140.3, 141.5, 143.1, 145.3, 147.8, 150.1, 153.1, 153.2, 156.1, 157.1, 159.8, 160.5, 162.8, 163.1, 165.6, 168.8, 169.7, 174.0, 178.3, 178.3, 178.5, 178.7, 178.9, 181.9, 188.3, 191.3, 194.6, 212.6 ] } ] }, "vllm-kv-fp8-triton": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 10.75, "min": 10.58, "max": 10.92, "n": 2 }, "median_ttft_ms": { "median": 716.98, "min": 716.57, "max": 717.39, "n": 2 }, "p99_ttft_ms": { "median": 721.61, "min": 719.07, "max": 724.16, "n": 2 }, "median_tpot_ms": { "median": 101.79, "min": 101.79, "max": 101.8, "n": 2 }, "p99_tpot_ms": { "median": 101.83, "min": 101.82, "max": 101.84, "n": 2 }, "p99_itl_ms": { "median": 102.42, "min": 102.35, "max": 102.49, "n": 2 }, "median_itl_ms": { "median": 101.7, "min": 101.68, "max": 101.72, "n": 2 }, "goodput_rps": { "median": 0.03, "min": 0.03, "max": 0.03, "n": 2 }, "request_throughput": { "median": 0.08, "min": 0.08, "max": 0.09, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 210.0, "min": 210.0, "max": 210.0, "n": 2 }, "max_temp_c": 95.0, "thermal_share": 0.793 }, "ttft_quantiles_ms": [ 77.3, 77.3, 78.9, 78.9, 183.0, 186.5, 186.5, 187.8, 187.8, 188.6, 189.7, 189.7, 222.4, 251.3, 251.3, 467.0, 467.0, 478.6, 711.8, 711.8, 714.9, 715.7, 715.7, 716.4, 716.4, 716.4, 716.7, 716.7, 717.0, 717.0, 717.1, 717.2, 717.2, 717.3, 717.5, 717.5, 718.4, 718.4, 718.7, 718.8, 718.8, 718.9, 719.1, 719.1, 719.2, 719.2, 719.3, 719.9, 719.9, 720.0, 724.9 ], "tpot_quantiles_ms": [ 59.8, 59.8, 59.8, 59.8, 59.8, 59.8, 59.8, 59.8, 59.8, 59.8, 59.9, 59.9, 68.4, 70.4, 70.4, 81.8, 81.8, 85.3, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8, 101.8 ] }, { "x": 32.0, "output_throughput": { "median": 192.0, "min": 183.28, "max": 200.72, "n": 2 }, "median_ttft_ms": { "median": 4266.89, "min": 4256.41, "max": 4277.36, "n": 2 }, "p99_ttft_ms": { "median": 8427.18, "min": 7099.02, "max": 9755.34, "n": 2 }, "median_tpot_ms": { "median": 137.96, "min": 131.58, "max": 144.34, "n": 2 }, "p99_tpot_ms": { "median": 197.87, "min": 191.31, "max": 204.44, "n": 2 }, "p99_itl_ms": { "median": 1671.77, "min": 1602.73, "max": 1740.82, "n": 2 }, "median_itl_ms": { "median": 86.18, "min": 85.62, "max": 86.73, "n": 2 }, "goodput_rps": { "median": 0.22, "min": 0.21, "max": 0.23, "n": 2 }, "request_throughput": { "median": 1.5, "min": 1.43, "max": 1.57, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 480.0, "min": 480.0, "max": 480.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.711 }, "ttft_quantiles_ms": [ 154.6, 714.3, 715.7, 717.1, 718.1, 718.8, 720.4, 721.9, 732.3, 1119.7, 1358.7, 1943.1, 1956.5, 1979.1, 2084.4, 2594.2, 2618.4, 2774.8, 2826.4, 3423.9, 3453.6, 3733.5, 4138.6, 4232.2, 4233.4, 4255.7, 4316.4, 4504.3, 4738.1, 4778.5, 4812.4, 4849.7, 4864.4, 4939.0, 4960.8, 5087.3, 5197.8, 5256.5, 5264.6, 5273.9, 5643.3, 5925.6, 6207.0, 6341.0, 6524.4, 6904.9, 7012.8, 7100.1, 8199.1, 8949.3, 9755.7 ], "tpot_quantiles_ms": [ 81.3, 82.9, 84.5, 85.3, 88.3, 89.1, 89.1, 89.1, 89.9, 89.9, 89.9, 89.9, 93.7, 105.8, 108.4, 118.7, 120.8, 123.3, 125.0, 129.2, 130.5, 134.7, 137.3, 137.5, 139.0, 142.0, 144.2, 144.3, 144.4, 144.5, 146.0, 149.2, 149.4, 151.4, 152.8, 153.4, 156.8, 157.2, 161.3, 163.7, 173.7, 176.6, 180.0, 180.2, 181.4, 182.1, 185.7, 186.1, 191.3, 196.2, 210.2 ] } ] }, "vllm-lock-b": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 11.2, "min": 10.88, "max": 11.51, "n": 2 }, "median_ttft_ms": { "median": 554.3, "min": 402.46, "max": 706.14, "n": 2 }, "p99_ttft_ms": { "median": 710.27, "min": 710.0, "max": 710.53, "n": 2 }, "median_tpot_ms": { "median": 99.03, "min": 93.7, "max": 104.35, "n": 2 }, "p99_tpot_ms": { "median": 104.4, "min": 104.4, "max": 104.41, "n": 2 }, "p99_itl_ms": { "median": 105.11, "min": 104.92, "max": 105.3, "n": 2 }, "median_itl_ms": { "median": 103.96, "min": 103.71, "max": 104.21, "n": 2 }, "goodput_rps": { "median": 0.04, "min": 0.03, "max": 0.04, "n": 2 }, "request_throughput": { "median": 0.09, "min": 0.09, "max": 0.09, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 210.0, "min": 210.0, "max": 210.0, "n": 2 }, "max_temp_c": 94.0, "thermal_share": 0.687 }, "ttft_quantiles_ms": [ 78.4, 78.4, 79.7, 79.7, 184.7, 185.9, 185.9, 185.9, 185.9, 185.9, 187.8, 187.8, 188.8, 190.0, 190.0, 191.6, 191.6, 193.0, 194.3, 194.3, 198.0, 225.1, 225.1, 554.5, 554.5, 579.8, 684.9, 684.9, 705.1, 705.1, 705.9, 706.4, 706.4, 706.4, 706.6, 706.6, 707.7, 707.7, 708.1, 708.2, 708.2, 708.2, 709.0, 709.0, 709.3, 709.3, 709.7, 710.1, 710.1, 710.2, 710.6 ], "tpot_quantiles_ms": [ 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.2, 61.2, 72.2, 83.6, 83.6, 101.4, 101.4, 103.8, 104.3, 104.3, 104.3, 104.3, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4, 104.4 ] }, { "x": 32.0, "output_throughput": { "median": 208.77, "min": 206.75, "max": 210.78, "n": 2 }, "median_ttft_ms": { "median": 3089.84, "min": 3088.89, "max": 3090.79, "n": 2 }, "p99_ttft_ms": { "median": 7107.31, "min": 6503.43, "max": 7711.18, "n": 2 }, "median_tpot_ms": { "median": 119.87, "min": 117.68, "max": 122.06, "n": 2 }, "p99_tpot_ms": { "median": 207.29, "min": 206.9, "max": 207.68, "n": 2 }, "p99_itl_ms": { "median": 1405.88, "min": 1389.58, "max": 1422.18, "n": 2 }, "median_itl_ms": { "median": 84.03, "min": 82.91, "max": 85.16, "n": 2 }, "goodput_rps": { "median": 0.13, "min": 0.1, "max": 0.15, "n": 2 }, "request_throughput": { "median": 1.63, "min": 1.62, "max": 1.65, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 881.25, "min": 862.5, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.5 }, "ttft_quantiles_ms": [ 120.1, 202.5, 216.9, 218.5, 400.3, 859.0, 860.6, 867.9, 869.4, 955.5, 973.7, 1512.4, 1527.7, 1604.0, 1606.4, 2079.4, 2176.9, 2179.9, 2249.5, 2619.3, 2681.6, 2827.1, 2846.4, 2861.0, 2907.1, 3089.8, 3124.2, 3202.3, 3208.8, 3228.4, 3293.7, 3482.6, 3485.1, 3872.1, 3880.4, 3933.1, 3935.1, 3945.6, 4149.0, 4184.3, 4267.8, 4286.4, 4342.5, 4710.8, 4809.1, 5228.5, 5574.9, 5672.0, 5987.7, 6213.8, 7981.9 ], "tpot_quantiles_ms": [ 93.0, 93.1, 93.2, 93.6, 93.7, 96.7, 96.7, 96.7, 97.2, 97.2, 101.2, 101.7, 101.7, 106.0, 106.1, 107.4, 108.3, 109.3, 109.4, 111.5, 113.2, 113.9, 114.7, 116.0, 117.7, 118.5, 119.8, 127.9, 129.0, 135.1, 137.9, 139.3, 144.3, 146.2, 146.8, 148.6, 149.4, 152.1, 154.4, 156.8, 157.7, 159.7, 162.6, 163.2, 168.4, 169.7, 172.9, 179.6, 193.8, 206.9, 220.9 ] } ] }, "vllm-lock-eagle3-k1": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 22.38, "min": 21.71, "max": 23.06, "n": 2 }, "median_ttft_ms": { "median": 240.02, "min": 234.08, "max": 245.95, "n": 2 }, "p99_ttft_ms": { "median": 662.76, "min": 619.22, "max": 706.29, "n": 2 }, "median_tpot_ms": { "median": 39.4, "min": 39.02, "max": 39.79, "n": 2 }, "p99_tpot_ms": { "median": 75.92, "min": 69.23, "max": 82.62, "n": 2 }, "p99_itl_ms": { "median": 99.64, "min": 84.47, "max": 114.8, "n": 2 }, "median_itl_ms": { "median": 68.32, "min": 68.32, "max": 68.32, "n": 2 }, "goodput_rps": { "median": 0.16, "min": 0.15, "max": 0.18, "n": 2 }, "request_throughput": { "median": 0.17, "min": 0.17, "max": 0.18, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.231 }, "ttft_quantiles_ms": [ 84.2, 84.2, 85.2, 85.2, 183.2, 183.6, 183.6, 183.7, 183.7, 183.9, 184.8, 184.8, 186.2, 187.5, 187.5, 189.3, 189.3, 189.6, 189.9, 189.9, 223.8, 228.3, 228.3, 244.3, 244.3, 245.4, 245.4, 245.4, 245.7, 245.7, 246.2, 246.2, 246.2, 247.3, 247.5, 247.5, 249.9, 249.9, 251.3, 251.9, 251.9, 260.2, 386.6, 386.6, 515.2, 515.2, 571.0, 637.6, 637.6, 704.8, 706.6 ], "tpot_quantiles_ms": [ 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.3, 68.4, 68.4, 68.4, 68.4, 68.4, 68.4, 68.4, 68.6, 80.2, 80.2, 80.8, 80.8, 80.9, 84.2, 84.2, 105.8, 111.7 ] }, { "x": 32.0, "output_throughput": { "median": 271.11, "min": 261.21, "max": 281.01, "n": 2 }, "median_ttft_ms": { "median": 1004.38, "min": 929.24, "max": 1079.52, "n": 2 }, "p99_ttft_ms": { "median": 3093.23, "min": 3016.36, "max": 3170.09, "n": 2 }, "median_tpot_ms": { "median": 100.43, "min": 94.09, "max": 106.78, "n": 2 }, "p99_tpot_ms": { "median": 158.56, "min": 150.07, "max": 167.05, "n": 2 }, "p99_itl_ms": { "median": 817.83, "min": 784.12, "max": 851.54, "n": 2 }, "median_itl_ms": { "median": 106.23, "min": 102.4, "max": 110.06, "n": 2 }, "goodput_rps": { "median": 0.01, "min": 0.0, "max": 0.02, "n": 2 }, "request_throughput": { "median": 2.12, "min": 2.04, "max": 2.2, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.447 }, "ttft_quantiles_ms": [ 126.1, 234.5, 285.0, 286.5, 438.0, 450.8, 480.9, 534.2, 593.5, 603.5, 668.8, 698.5, 731.0, 748.3, 779.1, 803.8, 890.3, 891.4, 892.9, 899.0, 903.3, 928.3, 944.9, 948.6, 1006.5, 1039.8, 1059.2, 1071.9, 1125.9, 1176.2, 1205.8, 1215.7, 1223.4, 1281.8, 1421.5, 1431.8, 1530.4, 1569.1, 1628.1, 1696.6, 1755.9, 1863.2, 1874.1, 1908.1, 2023.5, 2064.7, 2259.4, 2277.8, 2321.9, 3015.5, 3250.3 ], "tpot_quantiles_ms": [ 78.9, 112.4, 122.7, 127.1, 131.5, 133.0, 137.8, 138.6, 139.6, 142.4, 145.2, 147.5, 150.1, 152.1, 154.5, 155.8, 156.1, 157.9, 159.8, 162.7, 163.9, 164.9, 166.9, 168.3, 170.3, 171.0, 172.3, 174.2, 176.5, 179.1, 180.5, 185.2, 188.3, 190.1, 195.6, 198.6, 201.7, 206.1, 207.7, 211.3, 215.1, 223.7, 228.2, 233.5, 243.7, 249.6, 254.7, 259.8, 271.0, 277.0, 307.4 ] } ] }, "vllm-lock-eagle3-k3": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 31.31, "min": 31.19, "max": 31.44, "n": 2 }, "median_ttft_ms": { "median": 255.08, "min": 254.13, "max": 256.03, "n": 2 }, "p99_ttft_ms": { "median": 276.03, "min": 273.38, "max": 278.68, "n": 2 }, "median_tpot_ms": { "median": 27.81, "min": 26.8, "max": 28.83, "n": 2 }, "p99_tpot_ms": { "median": 60.78, "min": 58.65, "max": 62.92, "n": 2 }, "p99_itl_ms": { "median": 77.56, "min": 77.51, "max": 77.62, "n": 2 }, "median_itl_ms": { "median": 73.88, "min": 73.88, "max": 73.88, "n": 2 }, "goodput_rps": { "median": 0.24, "min": 0.24, "max": 0.25, "n": 2 }, "request_throughput": { "median": 0.24, "min": 0.24, "max": 0.25, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 89.0, "thermal_share": 0.0 }, "ttft_quantiles_ms": [ 83.3, 83.3, 85.9, 85.9, 184.2, 185.4, 185.4, 186.5, 186.5, 186.7, 186.8, 186.8, 187.5, 187.8, 187.8, 189.7, 189.7, 222.7, 251.8, 251.8, 253.3, 253.4, 253.4, 253.7, 253.7, 254.6, 254.8, 254.8, 255.7, 255.7, 256.2, 256.4, 256.4, 256.5, 256.9, 256.9, 258.2, 258.2, 258.3, 258.8, 258.8, 259.0, 259.5, 259.5, 260.0, 260.0, 262.2, 263.5, 263.5, 275.7, 281.4 ], "tpot_quantiles_ms": [ 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 73.9, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.0, 74.1, 74.1, 74.1, 74.1 ] }, { "x": 32.0, "output_throughput": { "median": 333.64, "min": 327.82, "max": 339.45, "n": 2 }, "median_ttft_ms": { "median": 890.48, "min": 838.97, "max": 941.99, "n": 2 }, "p99_ttft_ms": { "median": 3254.86, "min": 3233.28, "max": 3276.44, "n": 2 }, "median_tpot_ms": { "median": 77.84, "min": 76.49, "max": 79.18, "n": 2 }, "p99_tpot_ms": { "median": 156.61, "min": 156.5, "max": 156.72, "n": 2 }, "p99_itl_ms": { "median": 771.47, "min": 763.4, "max": 779.55, "n": 2 }, "median_itl_ms": { "median": 133.48, "min": 133.04, "max": 133.92, "n": 2 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 2 }, "request_throughput": { "median": 2.61, "min": 2.56, "max": 2.65, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.352 }, "ttft_quantiles_ms": [ 124.2, 272.9, 274.6, 277.6, 279.4, 479.8, 482.2, 488.1, 621.8, 625.1, 629.0, 633.3, 641.1, 662.7, 717.1, 770.7, 774.3, 778.2, 780.8, 782.6, 797.2, 839.1, 869.8, 896.6, 933.1, 935.6, 939.4, 940.5, 941.3, 1040.8, 1088.5, 1120.1, 1128.0, 1141.9, 1153.3, 1196.1, 1208.0, 1348.5, 1366.3, 1394.9, 1434.3, 1483.5, 1567.5, 1629.9, 1635.0, 1756.9, 1950.1, 2327.3, 2332.3, 3031.5, 3367.9 ], "tpot_quantiles_ms": [ 107.0, 128.8, 137.9, 152.2, 159.7, 165.8, 173.1, 175.9, 180.5, 182.9, 185.3, 185.9, 190.9, 193.6, 196.8, 199.4, 201.7, 202.7, 203.5, 204.4, 206.0, 206.7, 208.1, 209.0, 210.5, 212.0, 213.9, 215.7, 219.5, 222.4, 224.6, 226.3, 229.4, 235.6, 241.8, 243.9, 248.5, 256.6, 261.7, 270.2, 272.1, 276.8, 286.3, 295.3, 298.0, 300.7, 305.8, 317.8, 324.4, 330.0, 357.1 ] } ] }, "vllm-lock-eagle3-k5": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 31.73, "min": 30.61, "max": 32.85, "n": 2 }, "median_ttft_ms": { "median": 266.79, "min": 266.62, "max": 266.96, "n": 2 }, "p99_ttft_ms": { "median": 340.63, "min": 274.89, "max": 406.37, "n": 2 }, "median_tpot_ms": { "median": 26.21, "min": 25.11, "max": 27.3, "n": 2 }, "p99_tpot_ms": { "median": 78.95, "min": 63.68, "max": 94.22, "n": 2 }, "p99_itl_ms": { "median": 109.63, "min": 83.27, "max": 135.99, "n": 2 }, "median_itl_ms": { "median": 79.6, "min": 79.6, "max": 79.61, "n": 2 }, "goodput_rps": { "median": 0.24, "min": 0.22, "max": 0.26, "n": 2 }, "request_throughput": { "median": 0.25, "min": 0.24, "max": 0.26, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 89.0, "thermal_share": 0.136 }, "ttft_quantiles_ms": [ 84.2, 84.2, 84.9, 84.9, 185.0, 188.0, 188.0, 188.0, 188.0, 188.8, 189.6, 189.6, 190.4, 191.1, 191.1, 192.2, 192.2, 223.4, 265.6, 265.6, 265.6, 266.5, 266.5, 266.6, 266.6, 266.8, 266.8, 266.8, 267.2, 267.2, 267.7, 267.7, 267.7, 268.1, 268.8, 268.8, 268.8, 268.8, 269.1, 269.7, 269.7, 272.3, 273.3, 273.3, 275.2, 275.2, 285.2, 286.5, 286.5, 292.9, 426.4 ], "tpot_quantiles_ms": [ 79.6, 79.6, 79.6, 79.6, 79.6, 79.6, 79.6, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.7, 79.8, 79.8, 79.8, 79.8, 79.8, 79.8, 79.8, 79.8, 79.8, 79.8, 80.1, 113.2 ] }, { "x": 32.0, "output_throughput": { "median": 289.47, "min": 280.42, "max": 298.53, "n": 2 }, "median_ttft_ms": { "median": 1071.43, "min": 1059.7, "max": 1083.17, "n": 2 }, "p99_ttft_ms": { "median": 3356.6, "min": 3345.87, "max": 3367.33, "n": 2 }, "median_tpot_ms": { "median": 83.72, "min": 78.73, "max": 88.72, "n": 2 }, "p99_tpot_ms": { "median": 210.29, "min": 196.25, "max": 224.33, "n": 2 }, "p99_itl_ms": { "median": 915.23, "min": 835.83, "max": 994.64, "n": 2 }, "median_itl_ms": { "median": 227.96, "min": 218.39, "max": 237.54, "n": 2 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 2 }, "request_throughput": { "median": 2.26, "min": 2.19, "max": 2.33, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 885.0, "min": 885.0, "max": 885.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.432 }, "ttft_quantiles_ms": [ 125.3, 282.8, 284.4, 333.6, 335.1, 336.7, 573.1, 580.9, 640.0, 728.7, 737.0, 745.0, 792.8, 868.8, 889.0, 894.2, 895.9, 901.5, 905.8, 918.9, 956.2, 971.7, 1005.8, 1040.5, 1054.5, 1080.5, 1083.7, 1126.9, 1170.8, 1220.2, 1305.8, 1313.3, 1339.0, 1375.4, 1435.6, 1472.2, 1575.7, 1625.4, 1655.4, 1705.8, 1841.3, 1852.8, 1906.4, 1957.7, 1984.0, 2341.5, 2363.1, 2411.6, 2749.7, 3119.8, 3477.2 ], "tpot_quantiles_ms": [ 128.4, 167.9, 181.4, 192.4, 200.1, 210.4, 220.1, 223.8, 227.9, 230.9, 236.2, 238.3, 240.7, 245.0, 247.2, 248.6, 250.0, 251.7, 253.2, 255.1, 257.5, 260.4, 262.7, 267.5, 271.1, 272.4, 273.8, 276.4, 280.5, 285.3, 289.4, 295.0, 308.3, 315.9, 325.6, 341.5, 347.6, 349.8, 364.5, 371.9, 382.9, 386.5, 391.6, 405.3, 407.3, 420.2, 434.3, 476.1, 499.6, 507.6, 581.2 ] } ] }, "vllm-lock-fp8": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 25.41, "min": 24.51, "max": 26.31, "n": 2 }, "median_ttft_ms": { "median": 238.62, "min": 237.74, "max": 239.5, "n": 2 }, "p99_ttft_ms": { "median": 644.05, "min": 559.03, "max": 729.07, "n": 2 }, "median_tpot_ms": { "median": 35.25, "min": 35.25, "max": 35.25, "n": 2 }, "p99_tpot_ms": { "median": 47.57, "min": 43.12, "max": 52.02, "n": 2 }, "p99_itl_ms": { "median": 48.47, "min": 44.54, "max": 52.4, "n": 2 }, "median_itl_ms": { "median": 35.26, "min": 35.25, "max": 35.26, "n": 2 }, "goodput_rps": { "median": 0.2, "min": 0.19, "max": 0.21, "n": 2 }, "request_throughput": { "median": 0.2, "min": 0.19, "max": 0.21, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.226 }, "ttft_quantiles_ms": [ 49.7, 49.7, 50.5, 50.5, 235.1, 235.4, 235.4, 235.7, 235.7, 236.0, 236.0, 236.0, 236.3, 236.4, 236.4, 236.6, 236.6, 237.0, 237.0, 237.0, 237.1, 237.6, 237.6, 237.7, 237.7, 237.8, 237.9, 237.9, 238.0, 238.0, 238.8, 240.2, 240.2, 240.4, 241.4, 241.4, 254.5, 254.5, 257.8, 268.6, 268.6, 278.7, 289.3, 289.3, 437.4, 437.4, 580.5, 703.2, 703.2, 716.0, 731.4 ], "tpot_quantiles_ms": [ 35.1, 35.1, 35.1, 35.1, 35.1, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.2, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 35.3, 38.9, 38.9, 42.9, 42.9, 43.2, 52.0, 52.0, 52.0, 52.0 ] }, { "x": 32.0, "output_throughput": { "median": 284.68, "min": 272.31, "max": 297.06, "n": 2 }, "median_ttft_ms": { "median": 2861.79, "min": 2700.73, "max": 3022.84, "n": 2 }, "p99_ttft_ms": { "median": 6431.78, "min": 6324.15, "max": 6539.4, "n": 2 }, "median_tpot_ms": { "median": 83.82, "min": 81.53, "max": 86.11, "n": 2 }, "p99_tpot_ms": { "median": 156.31, "min": 140.74, "max": 171.88, "n": 2 }, "p99_itl_ms": { "median": 1143.45, "min": 1027.02, "max": 1259.88, "n": 2 }, "median_itl_ms": { "median": 49.48, "min": 49.3, "max": 49.65, "n": 2 }, "goodput_rps": { "median": 0.3, "min": 0.27, "max": 0.33, "n": 2 }, "request_throughput": { "median": 2.22, "min": 2.13, "max": 2.32, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 900.0, "min": 900.0, "max": 900.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.349 }, "ttft_quantiles_ms": [ 94.5, 158.5, 194.6, 196.2, 354.7, 412.2, 988.2, 989.7, 998.8, 1024.8, 1135.9, 1278.1, 1621.8, 1803.1, 1823.2, 1963.8, 1966.2, 2376.4, 2458.7, 2549.3, 2639.1, 2663.9, 2685.0, 2700.9, 2790.8, 2837.1, 2865.0, 3023.4, 3126.6, 3330.2, 3517.1, 3544.4, 3579.8, 3687.2, 3731.9, 4127.2, 4179.7, 4192.7, 4406.5, 4419.2, 4720.6, 4761.3, 4901.2, 4907.2, 4952.2, 5004.1, 5011.0, 5022.6, 5297.4, 6182.7, 6691.8 ], "tpot_quantiles_ms": [ 54.9, 58.9, 63.4, 63.5, 64.4, 64.7, 68.1, 68.1, 69.3, 69.3, 71.1, 74.2, 74.2, 75.4, 75.5, 80.3, 80.4, 80.4, 80.5, 80.5, 80.6, 81.4, 81.5, 81.7, 83.4, 83.5, 85.8, 86.0, 86.0, 86.1, 89.5, 89.7, 92.1, 92.2, 94.1, 94.4, 95.5, 98.8, 99.0, 99.8, 100.7, 103.5, 107.5, 108.1, 112.4, 119.5, 124.9, 137.4, 140.7, 162.2, 181.3 ] } ] }, "vllm-lock-small-pool": { "kv_pool_tokens": 6560, "points": [ { "x": 1.0, "output_throughput": { "median": 11.6, "min": 10.9, "max": 12.3, "n": 2 }, "median_ttft_ms": { "median": 449.14, "min": 208.97, "max": 689.3, "n": 2 }, "p99_ttft_ms": { "median": 705.83, "min": 703.92, "max": 707.74, "n": 2 }, "median_tpot_ms": { "median": 82.61, "min": 61.12, "max": 104.09, "n": 2 }, "p99_tpot_ms": { "median": 104.1, "min": 104.1, "max": 104.11, "n": 2 }, "p99_itl_ms": { "median": 104.71, "min": 104.65, "max": 104.77, "n": 2 }, "median_itl_ms": { "median": 82.66, "min": 61.35, "max": 103.98, "n": 2 }, "goodput_rps": { "median": 0.05, "min": 0.03, "max": 0.06, "n": 2 }, "request_throughput": { "median": 0.09, "min": 0.09, "max": 0.1, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 281.25, "min": 210.0, "max": 352.5, "n": 2 }, "max_temp_c": 94.0, "thermal_share": 0.682 }, "ttft_quantiles_ms": [ 79.5, 79.5, 81.4, 81.4, 183.8, 183.9, 183.9, 185.1, 185.1, 186.2, 187.1, 187.1, 187.3, 188.1, 188.1, 188.6, 188.6, 188.7, 190.5, 190.5, 191.2, 195.5, 195.5, 222.5, 222.5, 246.1, 565.3, 565.3, 677.6, 677.6, 697.5, 700.6, 700.6, 701.0, 701.7, 701.7, 701.8, 701.8, 701.9, 702.0, 702.0, 703.1, 704.1, 704.1, 704.1, 704.1, 704.9, 706.3, 706.3, 707.4, 707.8 ], "tpot_quantiles_ms": [ 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 61.1, 71.3, 71.3, 88.5, 103.0, 103.0, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1, 104.1 ] }, { "x": 32.0, "output_throughput": { "median": 80.44, "min": 79.83, "max": 81.05, "n": 2 }, "median_ttft_ms": { "median": 31567.67, "min": 31432.85, "max": 31702.48, "n": 2 }, "p99_ttft_ms": { "median": 48028.04, "min": 47880.42, "max": 48175.66, "n": 2 }, "median_tpot_ms": { "median": 120.56, "min": 118.87, "max": 122.25, "n": 2 }, "p99_tpot_ms": { "median": 248.17, "min": 245.5, "max": 250.84, "n": 2 }, "p99_itl_ms": { "median": 767.86, "min": 764.7, "max": 771.01, "n": 2 }, "median_itl_ms": { "median": 105.53, "min": 104.73, "max": 106.32, "n": 2 }, "goodput_rps": { "median": 0.02, "min": 0.02, "max": 0.02, "n": 2 }, "request_throughput": { "median": 0.63, "min": 0.62, "max": 0.63, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 247.5, "min": 240.0, "max": 255.0, "n": 2 }, "max_temp_c": 89.0, "thermal_share": 0.711 }, "ttft_quantiles_ms": [ 125.5, 762.7, 1403.9, 1407.1, 1930.3, 11306.7, 11950.2, 11952.7, 12476.8, 15523.0, 19017.9, 21611.0, 22173.0, 22248.9, 22515.4, 22817.9, 22892.4, 23313.9, 23383.8, 24334.0, 25373.8, 25748.0, 27891.9, 29562.6, 30746.5, 31466.2, 32038.7, 33153.0, 35526.0, 37557.1, 38977.3, 40356.5, 41128.1, 42286.0, 42388.2, 42905.3, 43702.1, 43805.8, 44023.7, 44340.7, 44451.7, 44644.4, 44933.8, 45370.5, 45974.5, 46097.3, 46205.2, 46327.5, 46588.6, 47854.1, 48294.7 ], "tpot_quantiles_ms": [ 71.3, 71.9, 72.9, 72.9, 72.9, 73.2, 73.9, 74.4, 74.8, 75.1, 76.6, 77.4, 79.6, 80.5, 84.4, 98.4, 107.3, 109.7, 110.8, 113.5, 114.4, 116.1, 118.9, 120.2, 120.6, 120.9, 121.8, 122.7, 126.4, 127.0, 128.1, 128.5, 128.5, 128.8, 130.7, 131.4, 131.7, 134.2, 136.1, 137.5, 137.6, 140.9, 145.0, 148.1, 149.1, 157.1, 173.9, 182.7, 214.3, 235.5, 257.2 ] } ] }, "vllm-eagle3": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 28.49, "min": 28.1, "max": 35.9, "n": 3 }, "median_ttft_ms": { "median": 264.8, "min": 261.81, "max": 267.13, "n": 3 }, "p99_ttft_ms": { "median": 657.93, "min": 563.49, "max": 708.0, "n": 3 }, "median_tpot_ms": { "median": 27.6, "min": 22.49, "max": 29.03, "n": 3 }, "p99_tpot_ms": { "median": 71.34, "min": 39.07, "max": 82.97, "n": 3 }, "p99_itl_ms": { "median": 96.16, "min": 80.7, "max": 107.43, "n": 3 }, "median_itl_ms": { "median": 72.53, "min": 72.36, "max": 72.6, "n": 3 }, "goodput_rps": { "median": 0.22, "min": 0.19, "max": 0.28, "n": 3 }, "request_throughput": { "median": 0.22, "min": 0.22, "max": 0.28, "n": 3 }, "completed": [ 16, 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1395.0, "min": 1327.5, "max": 1410.0, "n": 3 }, "max_temp_c": 89.0, "thermal_share": 0.468 }, "ttft_quantiles_ms": [ 81.0, 81.0, 82.5, 82.8, 180.8, 188.2, 188.9, 189.5, 190.0, 190.1, 190.9, 193.0, 193.7, 198.7, 223.2, 256.8, 260.3, 260.3, 260.4, 260.8, 261.5, 261.8, 262.1, 262.3, 263.1, 263.3, 263.5, 266.3, 270.7, 274.8, 276.8, 308.2, 385.2, 435.3, 435.3, 480.9, 481.3, 482.8, 553.7, 556.8, 557.3, 564.6, 579.3, 644.7, 647.9, 648.3, 653.6, 656.5, 658.7, 663.1, 715.9 ], "tpot_quantiles_ms": [ 72.0, 72.0, 72.0, 72.1, 72.1, 72.1, 72.2, 72.2, 72.2, 72.2, 72.2, 72.2, 72.2, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.3, 72.4, 72.5, 72.6, 73.8, 73.8, 73.9, 75.4, 75.4, 78.1, 79.3, 79.5, 79.8, 79.9, 80.0, 80.1, 80.5, 89.9, 89.9, 90.7, 91.4, 91.7, 99.4, 100.2, 100.4 ] }, { "x": 32.0, "output_throughput": { "median": 302.47, "min": 290.57, "max": 315.4, "n": 3 }, "median_ttft_ms": { "median": 959.76, "min": 921.92, "max": 1044.14, "n": 3 }, "p99_ttft_ms": { "median": 2889.47, "min": 2873.4, "max": 3093.4, "n": 3 }, "median_tpot_ms": { "median": 83.26, "min": 76.56, "max": 88.85, "n": 3 }, "p99_tpot_ms": { "median": 193.29, "min": 172.46, "max": 196.25, "n": 3 }, "p99_itl_ms": { "median": 850.53, "min": 693.49, "max": 958.45, "n": 3 }, "median_itl_ms": { "median": 158.45, "min": 145.22, "max": 162.09, "n": 3 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 3 }, "request_throughput": { "median": 2.36, "min": 2.27, "max": 2.46, "n": 3 }, "completed": [ 128, 128, 128 ], "gpu": { "median_sm_mhz": { "median": 862.5, "min": 742.5, "max": 892.5, "n": 3 }, "max_temp_c": 88.0, "thermal_share": 0.526 }, "ttft_quantiles_ms": [ 120.5, 278.3, 319.8, 322.0, 476.2, 483.1, 490.0, 586.8, 629.9, 634.4, 654.6, 709.6, 714.2, 716.3, 723.6, 760.9, 778.5, 798.8, 859.5, 886.3, 906.4, 918.5, 926.6, 937.5, 953.5, 996.1, 1012.7, 1026.1, 1052.3, 1069.0, 1131.9, 1162.8, 1207.9, 1229.6, 1272.0, 1305.2, 1359.1, 1440.1, 1515.9, 1532.4, 1544.4, 1563.1, 1580.2, 1597.7, 1645.8, 1895.9, 2005.1, 2228.7, 2356.7, 2872.6, 3180.5 ], "tpot_quantiles_ms": [ 110.2, 131.2, 146.0, 155.0, 161.8, 169.7, 172.8, 176.6, 180.5, 182.8, 185.7, 191.3, 192.9, 194.6, 199.0, 201.6, 203.2, 205.2, 207.0, 209.1, 213.5, 215.8, 219.9, 222.9, 225.2, 227.0, 231.0, 236.7, 241.8, 247.0, 255.3, 260.0, 265.6, 268.8, 276.2, 279.7, 284.0, 288.3, 292.6, 295.0, 298.5, 306.7, 315.1, 321.7, 337.4, 348.1, 359.6, 379.2, 386.6, 399.7, 447.5 ] } ] }, "vllm-eagle3-k5": { "kv_pool_tokens": 24096, "points": [ { "x": 1.0, "output_throughput": { "median": 28.88, "min": 28.47, "max": 29.29, "n": 2 }, "median_ttft_ms": { "median": 285.77, "min": 282.86, "max": 288.67, "n": 2 }, "p99_ttft_ms": { "median": 682.09, "min": 677.19, "max": 686.98, "n": 2 }, "median_tpot_ms": { "median": 26.4, "min": 26.13, "max": 26.67, "n": 2 }, "p99_tpot_ms": { "median": 92.28, "min": 80.58, "max": 103.99, "n": 2 }, "p99_itl_ms": { "median": 124.43, "min": 111.48, "max": 137.38, "n": 2 }, "median_itl_ms": { "median": 78.13, "min": 78.13, "max": 78.14, "n": 2 }, "goodput_rps": { "median": 0.18, "min": 0.16, "max": 0.21, "n": 2 }, "request_throughput": { "median": 0.23, "min": 0.22, "max": 0.23, "n": 2 }, "completed": [ 16, 16 ], "gpu": { "median_sm_mhz": { "median": 1357.5, "min": 1320.0, "max": 1395.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.462 }, "ttft_quantiles_ms": [ 81.8, 81.8, 89.2, 89.2, 190.2, 192.2, 192.2, 195.6, 195.6, 200.1, 215.9, 215.9, 267.3, 269.0, 269.0, 270.4, 270.4, 271.1, 271.7, 271.7, 272.4, 273.4, 273.4, 273.6, 273.6, 274.4, 291.4, 291.4, 300.0, 300.0, 305.0, 311.3, 311.3, 414.0, 497.8, 497.8, 553.9, 553.9, 579.5, 587.9, 587.9, 594.4, 594.6, 594.6, 666.7, 666.7, 675.5, 685.7, 685.7, 687.2, 691.8 ], "tpot_quantiles_ms": [ 77.6, 77.6, 77.6, 77.6, 77.7, 77.8, 77.8, 77.8, 77.8, 77.8, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 77.9, 78.0, 78.0, 78.0, 78.0, 78.0, 78.3, 78.6, 78.6, 87.2, 87.5, 87.5, 88.1, 88.1, 88.8, 90.1, 90.1, 91.8, 103.9, 103.9, 104.4, 104.4, 104.8, 105.2, 105.2, 105.9, 127.0 ] }, { "x": 32.0, "output_throughput": { "median": 287.04, "min": 285.25, "max": 288.83, "n": 2 }, "median_ttft_ms": { "median": 1185.28, "min": 1142.79, "max": 1227.76, "n": 2 }, "p99_ttft_ms": { "median": 3175.38, "min": 3121.15, "max": 3229.61, "n": 2 }, "median_tpot_ms": { "median": 84.43, "min": 79.57, "max": 89.3, "n": 2 }, "p99_tpot_ms": { "median": 206.17, "min": 205.5, "max": 206.83, "n": 2 }, "p99_itl_ms": { "median": 1005.37, "min": 914.71, "max": 1096.03, "n": 2 }, "median_itl_ms": { "median": 207.04, "min": 206.64, "max": 207.43, "n": 2 }, "goodput_rps": { "median": 0.0, "min": 0.0, "max": 0.0, "n": 2 }, "request_throughput": { "median": 2.24, "min": 2.23, "max": 2.26, "n": 2 }, "completed": [ 128, 128 ], "gpu": { "median_sm_mhz": { "median": 742.5, "min": 735.0, "max": 750.0, "n": 2 }, "max_temp_c": 88.0, "thermal_share": 0.506 }, "ttft_quantiles_ms": [ 123.1, 272.1, 273.7, 319.4, 321.0, 322.8, 583.6, 589.9, 700.5, 723.8, 744.8, 798.7, 870.8, 878.5, 883.6, 885.4, 897.4, 907.3, 927.7, 940.7, 970.3, 1037.0, 1049.5, 1140.6, 1150.2, 1179.5, 1210.1, 1234.1, 1296.5, 1303.8, 1336.4, 1442.8, 1452.5, 1464.5, 1479.2, 1513.5, 1549.1, 1576.6, 1641.1, 1770.9, 1869.5, 1892.4, 2002.4, 2013.0, 2188.4, 2233.8, 2272.8, 2759.8, 2835.4, 2885.6, 3229.7 ], "tpot_quantiles_ms": [ 143.1, 172.0, 180.9, 188.4, 204.0, 210.9, 217.6, 219.6, 228.0, 229.9, 232.5, 235.8, 239.8, 242.5, 246.2, 248.7, 250.4, 256.0, 258.3, 260.8, 264.0, 265.7, 268.2, 271.4, 276.0, 280.0, 282.1, 291.1, 294.7, 299.4, 309.4, 314.6, 323.6, 333.4, 344.6, 349.5, 355.7, 360.7, 366.7, 373.2, 379.8, 383.8, 393.8, 400.2, 403.2, 416.4, 430.9, 440.2, 449.9, 478.6, 518.6 ] } ] } }, "rate": { "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 ] } ] }, "sglang": { "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 ] } ] } }, "prefix": { "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 ] } ] }, "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 ] } ] }, "sglang": { "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 ] } ] } }, "waves": { "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 ] } ] }, "sglang": { "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 ] } ] } } }