{ "label": "", "model": "sensenova-loop-nvfp4", "base_url": "http://127.0.0.1:30080/v1", "host": "tux", "time": "2026-10-30T17:21:08-0700", "config": { "base_url": "http://127.0.0.1:30080/v1", "model": "sensenova-loop-nvfp4", "concurrency": "1,2,4,8", "num_requests": 8, "warmup": 2, "input_tokens": 1024, "output_tokens": 256, "tokenizer": "upstream/SenseNova-Loop-9B-Preview/tokenizer.json", "allow_prefix_cache": false, "no_ignore_eos": false, "timeout": 600, "label": "", "out": "eval/runs/perf_sglang_nvfp4_v1_mtp.json", "prompt_text_tokens": 1024, "prompt_build": "tokenizer.json" }, "levels": [ { "concurrency": 1, "n_requests": 8, "n_ok": 8, "n_errors": 0, "errors": [], "wall_s": 167.7497703190893, "ttft_s": { "p50": 0.7368478225544095, "p90": 0.7410881330259145, "mean": 0.7100226980401203 }, "e2e_s": { "p50": 20.956795112695545, "p90": 22.442169491481035, "mean": 20.96854761627037 }, "decode_tok_s_per_req": { "p50": 12.574976715274474, "p90": 13.522581057360785, "mean": 12.629198610683233 }, "itl_s": { "p50": 0.07952795094609552, "p90": 0.08523689763364839, "mean": 0.07944462552721447 }, "output_tokens_mean": 256.0, "output_tokens_min": 256, "output_tokens_max": 256, "prompt_tokens_mean": 1051.0, "n_short_outputs": 0, "aggregate_output_tok_s": 12.208660531125302, "requests_per_s": 0.04769008019970821, "usage_reported": true }, { "concurrency": 2, "n_requests": 8, "n_ok": 8, "n_errors": 0, "errors": [], "wall_s": 82.62508355267346, "ttft_s": { "p50": 0.8884775410406291, "p90": 1.168848348222673, "mean": 0.9600589442998171 }, "e2e_s": { "p50": 20.495888442732394, "p90": 23.237484936509283, "mean": 20.539230286842212 }, "decode_tok_s_per_req": { "p50": 13.113690417487145, "p90": 15.40404216440473, "mean": 13.185971019742919 }, "itl_s": { "p50": 0.07629031224687602, "p90": 0.08763312555082581, "mean": 0.0767804972402861 }, "output_tokens_mean": 256.0, "output_tokens_min": 256, "output_tokens_max": 256, "prompt_tokens_mean": 1051.0, "n_short_outputs": 0, "aggregate_output_tok_s": 24.786661773169655, "requests_per_s": 0.09682289755144396, "usage_reported": true }, { "concurrency": 4, "n_requests": 8, "n_ok": 8, "n_errors": 0, "errors": [], "wall_s": 48.18132451828569, "ttft_s": { "p50": 1.4830374312587082, "p90": 2.075450401753187, "mean": 1.469463941641152 }, "e2e_s": { "p50": 22.7819732921198, "p90": 24.27312577702105, "mean": 22.48564152873587 }, "decode_tok_s_per_req": { "p50": 12.02420852876317, "p90": 13.446927005044758, "mean": 12.240232030320389 }, "itl_s": { "p50": 0.08323863003512516, "p90": 0.09169133495101157, "mean": 0.08241570999380202 }, "output_tokens_mean": 256.0, "output_tokens_min": 256, "output_tokens_max": 256, "prompt_tokens_mean": 1051.0, "n_short_outputs": 0, "aggregate_output_tok_s": 42.50609588830931, "requests_per_s": 0.16603943706370825, "usage_reported": true }, { "concurrency": 8, "n_requests": 8, "n_ok": 8, "n_errors": 0, "errors": [], "wall_s": 31.786477480083704, "ttft_s": { "p50": 5.049089141655713, "p90": 5.0829355555586515, "mean": 5.062669638660736 }, "e2e_s": { "p50": 25.993340300396085, "p90": 29.076695342641322, "mean": 26.93958732835017 }, "decode_tok_s_per_req": { "p50": 12.175041260252009, "p90": 12.51746500245328, "mean": 11.74804181220008 }, "itl_s": { "p50": 0.08213639315948182, "p90": 0.09391973348115297, "mean": 0.08579115118813135 }, "output_tokens_mean": 256.0, "output_tokens_min": 256, "output_tokens_max": 256, "prompt_tokens_mean": 1051.0, "n_short_outputs": 0, "aggregate_output_tok_s": 64.42991367266805, "requests_per_s": 0.25167935028385957, "usage_reported": true } ] }