Files
llm_speed_test_app/results/run_20260826-004823.json

1528 lines
44 KiB
JSON

{
"version": "1.0",
"timestamp": "2026-08-26T00:48:23.163258",
"run_id": "20260826-004823",
"environment": {
"python_version": "3.12.0",
"platform": "win32",
"model": "Q3.6-35B-A3B-Orig-Thi",
"backend": "openai"
},
"config": {
"cases_requested": [
"case_01_generation",
"case_08_concurrency"
],
"repeats": 1,
"skip_heavy": false,
"total_cases": 2
},
"summary": {
"total_cases": 2,
"passed": 2,
"failed": 0,
"stopped": false,
"total_elapsed_seconds": 60.01
},
"results": [
{
"case_id": "case_01_generation",
"case_name": "纯文本生成基准",
"difficulty": "简单",
"status": "passed",
"raw_data": {
"case_id": "case_01_generation",
"name": "纯文本生成基准",
"description": "基础生成速度基准测试",
"difficulty": "简单",
"estimated_seconds": 5,
"results": [
{
"iteration": 1,
"ttft_ms": 1361.65,
"prefill_ms": 174.46,
"decode_speed_tok_s": 55.94,
"total_tokens": 331,
"prompt_tokens": 31,
"completion_tokens": 300,
"e2e_ms": 5672.64,
"elapsed_ms": 5672.64,
"response_length": 917
}
],
"prompt_used": "请用200字左右概述人工智能从诞生至今的关键发展阶段,每段不超过50字。"
},
"statistics": {
"count": 1,
"mean_ms": 5672.64,
"std_ms": 0.0,
"min_ms": 5672.64,
"max_ms": 5672.64,
"p50_ms": 5672.64,
"p95_ms": 5672.64,
"p99_ms": 5672.64
},
"inference_metrics": {
"ttft_ms_mean": 1361.65,
"ttft_ms_p95": 1361.65,
"prefill_ms_mean": 174.46,
"prefill_ms_p95": 174.46,
"e2e_ms_mean": 5672.64,
"e2e_ms_p95": 5672.64,
"decode_speed_tok_s_mean": 55.94,
"decode_speed_tok_s_p95": 55.94,
"total_tokens_mean": 331.0,
"total_tokens_p95": 331
},
"elapsed_seconds": 5.67
},
{
"case_id": "case_08_concurrency",
"case_name": "并发压力测试",
"difficulty": "压力",
"status": "passed",
"raw_data": {
"case_id": "case_08_concurrency",
"name": "并发压力测试",
"description": "高并发下的 TPS 保持率、P95/P99 延迟、错误率",
"difficulty": "压力",
"estimated_seconds": 30,
"results": [
{
"iteration": 1,
"ttft_ms": 60.74,
"prefill_ms": 60.74,
"decode_speed_tok_s": 35.36,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 5674.36,
"elapsed_ms": 5674.36,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 2,
"ttft_ms": 56.05,
"prefill_ms": 56.05,
"decode_speed_tok_s": 41.47,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 5264.59,
"elapsed_ms": 5264.59,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 3,
"ttft_ms": 66.16,
"prefill_ms": 66.16,
"decode_speed_tok_s": 38.66,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 6600.18,
"elapsed_ms": 6600.18,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 4,
"ttft_ms": 68.09,
"prefill_ms": 68.09,
"decode_speed_tok_s": 38.55,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 6216.05,
"elapsed_ms": 6216.05,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 5,
"ttft_ms": 68.2,
"prefill_ms": 68.2,
"decode_speed_tok_s": 38.74,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 7129.09,
"elapsed_ms": 7129.09,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 6,
"ttft_ms": 66.17,
"prefill_ms": 66.17,
"decode_speed_tok_s": 38.73,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 7458.59,
"elapsed_ms": 7458.59,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 7,
"ttft_ms": 68.2,
"prefill_ms": 68.2,
"decode_speed_tok_s": 38.84,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 8022.2,
"elapsed_ms": 8022.2,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 8,
"ttft_ms": 66.1,
"prefill_ms": 66.1,
"decode_speed_tok_s": 38.64,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 8373.14,
"elapsed_ms": 8373.14,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 9,
"ttft_ms": 68.25,
"prefill_ms": 68.25,
"decode_speed_tok_s": 38.64,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 8911.84,
"elapsed_ms": 8911.84,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 10,
"ttft_ms": 66.69,
"prefill_ms": 66.69,
"decode_speed_tok_s": 38.6,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 9254.79,
"elapsed_ms": 9254.79,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 11,
"ttft_ms": 68.54,
"prefill_ms": 68.54,
"decode_speed_tok_s": 38.71,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 9801.75,
"elapsed_ms": 9801.75,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 12,
"ttft_ms": 66.39,
"prefill_ms": 66.39,
"decode_speed_tok_s": 38.66,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 10199.93,
"elapsed_ms": 10199.93,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 13,
"ttft_ms": 68.4,
"prefill_ms": 68.4,
"decode_speed_tok_s": 38.74,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 10687.59,
"elapsed_ms": 10687.59,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 14,
"ttft_ms": 66.35,
"prefill_ms": 66.35,
"decode_speed_tok_s": 38.71,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 11063.96,
"elapsed_ms": 11063.96,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 15,
"ttft_ms": 68.24,
"prefill_ms": 68.24,
"decode_speed_tok_s": 38.75,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 11593.13,
"elapsed_ms": 11593.13,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 16,
"ttft_ms": 66.28,
"prefill_ms": 66.28,
"decode_speed_tok_s": 38.68,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 11955.55,
"elapsed_ms": 11955.55,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 17,
"ttft_ms": 68.29,
"prefill_ms": 68.29,
"decode_speed_tok_s": 38.66,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 12490.11,
"elapsed_ms": 12490.11,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 18,
"ttft_ms": 66.25,
"prefill_ms": 66.25,
"decode_speed_tok_s": 38.48,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 12881.69,
"elapsed_ms": 12881.69,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 19,
"ttft_ms": 69.41,
"prefill_ms": 69.41,
"decode_speed_tok_s": 38.7,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 13384.74,
"elapsed_ms": 13384.74,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 20,
"ttft_ms": 66.24,
"prefill_ms": 66.24,
"decode_speed_tok_s": 38.74,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 13780.65,
"elapsed_ms": 13780.65,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 21,
"ttft_ms": 68.29,
"prefill_ms": 68.29,
"decode_speed_tok_s": 38.82,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 14265.17,
"elapsed_ms": 14265.17,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 22,
"ttft_ms": 66.26,
"prefill_ms": 66.26,
"decode_speed_tok_s": 38.75,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 14663.09,
"elapsed_ms": 14663.09,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 23,
"ttft_ms": 68.27,
"prefill_ms": 68.27,
"decode_speed_tok_s": 38.74,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 15171.04,
"elapsed_ms": 15171.04,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 24,
"ttft_ms": 66.21,
"prefill_ms": 66.21,
"decode_speed_tok_s": 38.59,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 15530.12,
"elapsed_ms": 15530.12,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 25,
"ttft_ms": 68.57,
"prefill_ms": 68.57,
"decode_speed_tok_s": 38.71,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 16066.75,
"elapsed_ms": 16066.75,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 26,
"ttft_ms": 66.25,
"prefill_ms": 66.25,
"decode_speed_tok_s": 38.69,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 16431.91,
"elapsed_ms": 16431.91,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 27,
"ttft_ms": 68.22,
"prefill_ms": 68.22,
"decode_speed_tok_s": 38.73,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 16970.02,
"elapsed_ms": 16970.02,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 28,
"ttft_ms": 66.24,
"prefill_ms": 66.24,
"decode_speed_tok_s": 38.68,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 17361.72,
"elapsed_ms": 17361.72,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 29,
"ttft_ms": 68.68,
"prefill_ms": 68.68,
"decode_speed_tok_s": 38.87,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 17844.61,
"elapsed_ms": 17844.61,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 30,
"ttft_ms": 66.18,
"prefill_ms": 66.18,
"decode_speed_tok_s": 38.78,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 18225.63,
"elapsed_ms": 18225.63,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 31,
"ttft_ms": 68.19,
"prefill_ms": 68.19,
"decode_speed_tok_s": 38.74,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 18774.54,
"elapsed_ms": 18774.54,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 32,
"ttft_ms": 66.18,
"prefill_ms": 66.18,
"decode_speed_tok_s": 38.73,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 19127.95,
"elapsed_ms": 19127.95,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 33,
"ttft_ms": 68.14,
"prefill_ms": 68.14,
"decode_speed_tok_s": 38.58,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 19670.18,
"elapsed_ms": 19670.18,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 34,
"ttft_ms": 66.95,
"prefill_ms": 66.95,
"decode_speed_tok_s": 38.53,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 19937.24,
"elapsed_ms": 19937.24,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 35,
"ttft_ms": 68.24,
"prefill_ms": 68.24,
"decode_speed_tok_s": 38.54,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 20555.94,
"elapsed_ms": 20555.94,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 36,
"ttft_ms": 66.44,
"prefill_ms": 66.44,
"decode_speed_tok_s": 38.62,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 20937.87,
"elapsed_ms": 20937.87,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 37,
"ttft_ms": 68.19,
"prefill_ms": 68.19,
"decode_speed_tok_s": 38.76,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 21435.19,
"elapsed_ms": 21435.19,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 38,
"ttft_ms": 66.34,
"prefill_ms": 66.34,
"decode_speed_tok_s": 38.68,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 21786.48,
"elapsed_ms": 21786.48,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 39,
"ttft_ms": 68.17,
"prefill_ms": 68.17,
"decode_speed_tok_s": 38.9,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 20920.76,
"elapsed_ms": 20920.76,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 40,
"ttft_ms": 66.17,
"prefill_ms": 66.17,
"decode_speed_tok_s": 38.88,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 22682.65,
"elapsed_ms": 22682.65,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 41,
"ttft_ms": 68.1,
"prefill_ms": 68.1,
"decode_speed_tok_s": 38.73,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 23215.5,
"elapsed_ms": 23215.5,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 42,
"ttft_ms": 66.24,
"prefill_ms": 66.24,
"decode_speed_tok_s": 38.59,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 21973.18,
"elapsed_ms": 21973.18,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 43,
"ttft_ms": 68.27,
"prefill_ms": 68.27,
"decode_speed_tok_s": 38.88,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 22938.81,
"elapsed_ms": 22938.81,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 44,
"ttft_ms": 66.11,
"prefill_ms": 66.11,
"decode_speed_tok_s": 38.78,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 24501.4,
"elapsed_ms": 24501.4,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 45,
"ttft_ms": 68.22,
"prefill_ms": 68.22,
"decode_speed_tok_s": 38.85,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25002.77,
"elapsed_ms": 25002.77,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 46,
"ttft_ms": 66.24,
"prefill_ms": 66.24,
"decode_speed_tok_s": 38.85,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 22689.68,
"elapsed_ms": 22689.68,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 47,
"ttft_ms": 68.39,
"prefill_ms": 68.39,
"decode_speed_tok_s": 38.91,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 23393.4,
"elapsed_ms": 23393.4,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 48,
"ttft_ms": 66.18,
"prefill_ms": 66.18,
"decode_speed_tok_s": 38.85,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 23347.92,
"elapsed_ms": 23347.92,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 49,
"ttft_ms": 68.23,
"prefill_ms": 68.23,
"decode_speed_tok_s": 38.94,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 23569.98,
"elapsed_ms": 23569.98,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 50,
"ttft_ms": 66.13,
"prefill_ms": 66.13,
"decode_speed_tok_s": 38.87,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 24929.72,
"elapsed_ms": 24929.72,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 51,
"ttft_ms": 68.34,
"prefill_ms": 68.34,
"decode_speed_tok_s": 39.0,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25878.51,
"elapsed_ms": 25878.51,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 52,
"ttft_ms": 66.11,
"prefill_ms": 66.11,
"decode_speed_tok_s": 38.86,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 26206.96,
"elapsed_ms": 26206.96,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 53,
"ttft_ms": 68.05,
"prefill_ms": 68.05,
"decode_speed_tok_s": 38.85,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25135.12,
"elapsed_ms": 25135.12,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 54,
"ttft_ms": 66.89,
"prefill_ms": 66.89,
"decode_speed_tok_s": 38.93,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25209.91,
"elapsed_ms": 25209.91,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 55,
"ttft_ms": 68.05,
"prefill_ms": 68.05,
"decode_speed_tok_s": 38.88,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25457.93,
"elapsed_ms": 25457.93,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 56,
"ttft_ms": 66.14,
"prefill_ms": 66.14,
"decode_speed_tok_s": 38.8,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25833.38,
"elapsed_ms": 25833.38,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 57,
"ttft_ms": 68.05,
"prefill_ms": 68.05,
"decode_speed_tok_s": 38.79,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25070.91,
"elapsed_ms": 25070.91,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 58,
"ttft_ms": 66.22,
"prefill_ms": 66.22,
"decode_speed_tok_s": 38.69,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25137.48,
"elapsed_ms": 25137.48,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 59,
"ttft_ms": 69.17,
"prefill_ms": 69.17,
"decode_speed_tok_s": 38.94,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25730.69,
"elapsed_ms": 25730.69,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 60,
"ttft_ms": 66.08,
"prefill_ms": 66.08,
"decode_speed_tok_s": 38.88,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 25735.77,
"elapsed_ms": 25735.77,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 61,
"ttft_ms": 68.08,
"prefill_ms": 68.08,
"decode_speed_tok_s": 38.87,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 26371.2,
"elapsed_ms": 26371.2,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 62,
"ttft_ms": 66.16,
"prefill_ms": 66.16,
"decode_speed_tok_s": 38.79,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 26661.55,
"elapsed_ms": 26661.55,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 63,
"ttft_ms": 68.08,
"prefill_ms": 68.08,
"decode_speed_tok_s": 39.07,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 27349.81,
"elapsed_ms": 27349.81,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 64,
"ttft_ms": 66.06,
"prefill_ms": 66.06,
"decode_speed_tok_s": 38.94,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 27720.81,
"elapsed_ms": 27720.81,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 65,
"ttft_ms": 68.17,
"prefill_ms": 68.17,
"decode_speed_tok_s": 38.8,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 28072.64,
"elapsed_ms": 28072.64,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 66,
"ttft_ms": 66.53,
"prefill_ms": 66.53,
"decode_speed_tok_s": 38.67,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 28089.05,
"elapsed_ms": 28089.05,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 67,
"ttft_ms": 68.21,
"prefill_ms": 68.21,
"decode_speed_tok_s": 38.84,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 29111.18,
"elapsed_ms": 29111.18,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 68,
"ttft_ms": 66.26,
"prefill_ms": 66.26,
"decode_speed_tok_s": 38.89,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 29487.39,
"elapsed_ms": 29487.39,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 69,
"ttft_ms": 68.03,
"prefill_ms": 68.03,
"decode_speed_tok_s": 38.91,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 29608.49,
"elapsed_ms": 29608.49,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 70,
"ttft_ms": 66.09,
"prefill_ms": 66.09,
"decode_speed_tok_s": 38.85,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 30064.46,
"elapsed_ms": 30064.46,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 71,
"ttft_ms": 68.52,
"prefill_ms": 68.52,
"decode_speed_tok_s": 38.92,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 30537.86,
"elapsed_ms": 30537.86,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 72,
"ttft_ms": 66.55,
"prefill_ms": 66.55,
"decode_speed_tok_s": 38.83,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 30794.6,
"elapsed_ms": 30794.6,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 73,
"ttft_ms": 68.67,
"prefill_ms": 68.67,
"decode_speed_tok_s": 39.11,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 37482.39,
"elapsed_ms": 37482.39,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 74,
"ttft_ms": 65.99,
"prefill_ms": 65.99,
"decode_speed_tok_s": 39.02,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 37869.44,
"elapsed_ms": 37869.44,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 75,
"ttft_ms": 68.5,
"prefill_ms": 68.5,
"decode_speed_tok_s": 39.11,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 31796.22,
"elapsed_ms": 31796.22,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 76,
"ttft_ms": 66.16,
"prefill_ms": 66.16,
"decode_speed_tok_s": 39.08,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 38776.85,
"elapsed_ms": 38776.85,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 77,
"ttft_ms": 67.94,
"prefill_ms": 67.94,
"decode_speed_tok_s": 39.16,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 39278.77,
"elapsed_ms": 39278.77,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 78,
"ttft_ms": 66.11,
"prefill_ms": 66.11,
"decode_speed_tok_s": 39.11,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 39626.86,
"elapsed_ms": 39626.86,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 79,
"ttft_ms": 67.89,
"prefill_ms": 67.89,
"decode_speed_tok_s": 39.2,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 40151.1,
"elapsed_ms": 40151.1,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 80,
"ttft_ms": 66.11,
"prefill_ms": 66.11,
"decode_speed_tok_s": 39.07,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 40534.17,
"elapsed_ms": 40534.17,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 81,
"ttft_ms": 67.98,
"prefill_ms": 67.98,
"decode_speed_tok_s": 39.09,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 41029.96,
"elapsed_ms": 41029.96,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 82,
"ttft_ms": 66.09,
"prefill_ms": 66.09,
"decode_speed_tok_s": 39.08,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 41436.63,
"elapsed_ms": 41436.63,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 83,
"ttft_ms": 68.0,
"prefill_ms": 68.0,
"decode_speed_tok_s": 39.21,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 41670.92,
"elapsed_ms": 41670.92,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 84,
"ttft_ms": 66.05,
"prefill_ms": 66.05,
"decode_speed_tok_s": 39.16,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42300.91,
"elapsed_ms": 42300.91,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 85,
"ttft_ms": 67.92,
"prefill_ms": 67.92,
"decode_speed_tok_s": 39.18,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42285.07,
"elapsed_ms": 42285.07,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 86,
"ttft_ms": 66.05,
"prefill_ms": 66.05,
"decode_speed_tok_s": 38.99,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42387.23,
"elapsed_ms": 42387.23,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 87,
"ttft_ms": 68.57,
"prefill_ms": 68.57,
"decode_speed_tok_s": 39.08,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42706.42,
"elapsed_ms": 42706.42,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 88,
"ttft_ms": 66.11,
"prefill_ms": 66.11,
"decode_speed_tok_s": 38.96,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 44074.46,
"elapsed_ms": 44074.46,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 89,
"ttft_ms": 67.82,
"prefill_ms": 67.82,
"decode_speed_tok_s": 39.06,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 44576.86,
"elapsed_ms": 44576.86,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 90,
"ttft_ms": 66.05,
"prefill_ms": 66.05,
"decode_speed_tok_s": 39.07,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 44959.12,
"elapsed_ms": 44959.12,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 91,
"ttft_ms": 68.2,
"prefill_ms": 68.2,
"decode_speed_tok_s": 39.25,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 44887.1,
"elapsed_ms": 44887.1,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 92,
"ttft_ms": 65.99,
"prefill_ms": 65.99,
"decode_speed_tok_s": 39.27,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 45825.94,
"elapsed_ms": 45825.94,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 93,
"ttft_ms": 67.79,
"prefill_ms": 67.79,
"decode_speed_tok_s": 39.23,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 45953.13,
"elapsed_ms": 45953.13,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 94,
"ttft_ms": 66.02,
"prefill_ms": 66.02,
"decode_speed_tok_s": 39.17,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 41492.84,
"elapsed_ms": 41492.84,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 95,
"ttft_ms": 67.86,
"prefill_ms": 67.86,
"decode_speed_tok_s": 39.32,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42895.55,
"elapsed_ms": 42895.55,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 96,
"ttft_ms": 65.97,
"prefill_ms": 65.97,
"decode_speed_tok_s": 39.26,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42080.43,
"elapsed_ms": 42080.43,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 97,
"ttft_ms": 67.88,
"prefill_ms": 67.88,
"decode_speed_tok_s": 39.39,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42992.32,
"elapsed_ms": 42992.32,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 98,
"ttft_ms": 65.96,
"prefill_ms": 65.96,
"decode_speed_tok_s": 39.28,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 43592.29,
"elapsed_ms": 43592.29,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 99,
"ttft_ms": 67.75,
"prefill_ms": 67.75,
"decode_speed_tok_s": 39.2,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 42537.22,
"elapsed_ms": 42537.22,
"response_length": 91,
"batch": "concurrent",
"error": ""
},
{
"iteration": 100,
"ttft_ms": 68.47,
"prefill_ms": 68.47,
"decode_speed_tok_s": 46.89,
"total_tokens": 36,
"prompt_tokens": 4,
"completion_tokens": 32,
"e2e_ms": 44598.8,
"elapsed_ms": 44598.8,
"response_length": 91,
"batch": "concurrent",
"error": ""
}
],
"concurrency": 100,
"baseline_tps": 57.43,
"concurrent_tps": 38.95,
"tps_keep_rate_pct": 67.82,
"e2e_p50_ms": 25036.84,
"e2e_p95_ms": 44577.95,
"e2e_p99_ms": 45827.21,
"success_count": 100,
"failed_count": 0,
"error_rate_pct": 0.0
},
"statistics": {
"count": 100,
"mean_ms": 25637.3,
"std_ms": 11979.73,
"min_ms": 5264.59,
"max_ms": 45953.13,
"p50_ms": 25036.84,
"p95_ms": 44577.96,
"p99_ms": 45827.21
},
"inference_metrics": {
"ttft_ms_mean": 67.07,
"ttft_ms_p95": 68.57,
"prefill_ms_mean": 67.07,
"prefill_ms_p95": 68.57,
"e2e_ms_mean": 25637.3,
"e2e_ms_p95": 44577.96,
"decode_speed_tok_s_mean": 38.95,
"decode_speed_tok_s_p95": 39.27,
"total_tokens_mean": 36.0,
"total_tokens_p95": 36.0
},
"elapsed_seconds": 54.33
}
]
}