{"id":"run_xitx0BeMs3KHD2V7","owner":"anonymous-EVjB2S","publishedAt":"2026-08-30T04:10:39.209Z","manifest":{"schema_version":"1.0","kind":"run","client_run_id":"rps-148b89ee5443bba5f221b1eb77b009f44b436677","name":"gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 | baseline | 4096->1 | C2","notes":"Aggregate-only synthetic online serving measurement. It reports serving performance and does not evaluate model quality.","visibility":"public","license":"CC-BY-4.0","outcome":"completed","subject":{"type":"model","source":"huggingface","adapters":[],"revision":"0cc27958cefbbe231782ec8511de8c4eb5233348","identifier":"gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090","modalities":["text"],"architecture":"Qwen3_5ForConditionalGeneration","quantization":{"format":"NVFP4 W4A4","recipe":"gittensor-model-hub RTX 5090 target checkpoint","checkpoint":"NVIDIA ModelOpt"},"modifications":[],"revision_kind":"commit","parameter_count_total":27000000000,"parameter_count_active":27000000000},"environment":{"os":"Linux-6.8.0-136-generic-x86_64-with-glibc2.39","cpu":{"model":"x86_64"},"kernel":"6.8.0-136-generic","topology":"One server engine on one GPU.","accelerators":[{"count":1,"model":"NVIDIA GeForce RTX 5090","driver":"580.173.02","vendor":"NVIDIA","architecture":"Blackwell SM120","interconnect":"single device","memory_bytes_per_device":34190917632}],"driver_versions":{"nvidia":"580.173.02"},"host_architecture":"x86_64","system_memory_bytes":269585743872},"runtime":{"engine":"sglang","version":"0.0.0.dev1+g5f55db35e","extensions":{"runpile.server_variant":"baseline","runpile.auxiliary_models":[]},"parameters":{"server":{"dtype":"bfloat16","context_length":9216,"kv_cache_dtype":"fp8_e4m3","attention_backend":"flashinfer","cuda_graph_max_bs":8,"mem_fraction_static":0.9,"chunked_prefill_size":1024,"max_running_requests":8}},"parallelism":{"data":1,"expert":1,"tensor":1,"pipeline":1,"replicas":1},"launch_command":"CUDA_VISIBLE_DEVICES=0 sglang serve --model-path gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 --revision 0cc27958cefbbe231782ec8511de8c4eb5233348 --served-model-name gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 --port 9000 --host 127.0.0.1 --dtype bfloat16 --mem-fraction-static 0.9 --context-length 9216 --max-running-requests 8 --tp 1 --kv-cache-dtype fp8_e4m3 --chunked-prefill-size 1024 --attention-backend flashinfer --cuda-graph-max-bs 8 --trust-remote-code --reasoning-parser qwen3 --tool-call-parser qwen3_coder --mamba-full-memory-ratio 5.61 --mamba-radix-cache-strategy extra_buffer_lazy --mamba-ssm-dtype bfloat16 --max-mamba-cache-size 32 --cuda-graph-bs-prefill 64 128 256 512 1024 --enable-memory-saver --enable-hierarchical-cache --hicache-ratio 2 --hicache-write-policy write_through","build_identifier":"lmsysorg/sglang@sha256:616a3e97f45191af975896cfa644279096cb31bd408a071c2e99ca7209c3cafe","safe_environment":{"HF_HUB_OFFLINE":"1","CUDA_VISIBLE_DEVICES":"0","TOKENIZERS_PARALLELISM":"false"}},"workload":{"mode":"online","task":"text_generation","client":{"name":"sglang benchmark serving","backend":"sglang","version":"0.0.0.dev1+g5f55db35e","placement":"same-host"},"dataset":{"generator":"sglang-benchmark-random-token-ids-fixed-v1","synthetic":true,"generator_revision":"b62df3a992da3d4a1dc6f2f12ce2706bbeef61bc5ca0c03c917d7597606c3312","shared_prefix_tokens":0},"sampling":{"mode":"greedy","temperature":0},"stopping":{"ignore_eos":true,"max_output_tokens":1},"streaming":true,"tokenizer":{"revision":"0cc27958cefbbe231782ec8511de8c4eb5233348","identifier":"gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090"},"concurrency":2,"random_seed":0,"input_tokens":{"kind":"fixed","value":4096},"endpoint_type":"sglang-native-completions","output_tokens":{"kind":"fixed","value":1},"request_count":256,"arrival_process":"closed_loop","prefix_behavior":{"cache_state":"warm-host","cache_enabled":false,"shared_prefix_tokens":0},"warmup_request_count":16,"request_rate_per_second":null},"measurement":{"notes":"Aggregate-only output; TPOT/ITL are inapplicable for a one-output-token workload.","warmup":"16 unmeasured warmups per invocation after one readiness request.","harness":"sglang benchmark serving","version":"0.0.0.dev1+g5f55db35e","extensions":{"runpile.declared_warmup_count":16,"runpile.max_gpu_temperature_c":57,"runpile.preemption_log_mentions":2,"runpile.readiness_request_count":1,"runpile.controller_source_sha256":"b62df3a992da3d4a1dc6f2f12ce2706bbeef61bc5ca0c03c917d7597606c3312","runpile.independent_repetition_index":1},"repetitions":1,"timing_scope":"client-observed streaming request latency","benchmark_command":"sglang bench serving --backend sglang --base-url http://127.0.0.1:9000 --model gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 --served-model-name gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 --tokenizer gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090 --dataset-name random-ids --dataset-path <EXACT_SYNTHETIC_DATASET> --random-input-len 4096 --random-output-len 1 --random-range-ratio 1 --tokenize-prompt --num-prompts 256 --max-concurrency 2 --request-rate inf --seed 0 --temperature 0 --disable-tqdm --output-file <local-path> --warmup-requests 16 --tag runpile-rtx5090-qwen38-dflash2-concurrency-20260829-v5:qwen38-27b-nvfp4-sglang-target-only-tp1:baseline:serve-fixed-4k-1-categorization-v1:c2:r1","retry_failure_treatment":"Failed requests counted; no client retries.","client_overhead_included":true},"results":{"latency_ms":{"end_to_end":{"sample_count":256,"measurement_status":"measured"},"inter_token":{"p50":0,"p90":0,"p95":0,"p99":0,"mean":0,"stddev":0,"sample_count":256,"measurement_status":"measured"},"time_to_first_token":{"p50":536.1974616535008,"p90":595.4177745152265,"p95":596.1915662046522,"p99":596.5869590407237,"mean":530.5562410685525,"stddev":62.56015280725047,"sample_count":256,"measurement_status":"measured"},"time_per_output_token":{"p50":0,"p90":0,"p95":0,"p99":0,"mean":0,"stddev":0,"sample_count":256,"measurement_status":"measured"}},"throughput":{"requests_per_second":{"value":3.732773214291039,"sample_count":256,"measurement_status":"measured"},"input_tokens_per_second":{"value":15289.439085736096,"sample_count":256,"measurement_status":"measured"},"output_tokens_per_second":{"value":3.732773214291039,"sample_count":256,"measurement_status":"measured"}},"failed_requests":0,"duration_seconds":68.58171801595017,"total_input_tokens":1048576,"successful_requests":256,"total_output_tokens":256},"artifacts":[{"kind":"sglang_bench_serving_json","sha256":"dfa326fa93f103580e0bddd384499b8019122cefb3863a083a61ed3dcfed3fd6","license":"CC-BY-4.0","filename":"qwen38-27b-nvfp4-sglang-target-only-tp1__baseline__serve-fixed-4k-1-categorization-v1__c2__r1.json","media_type":"application/json","size_bytes":43300,"visibility":"private","client_artifact_id":"raw-sglang-serve","raw_requests_included":false}],"provenance":{"tool":"runpile HTTP importer","source":"import","tool_version":"1.0"},"warnings":["Review imported facts against the local model, hardware, launch command, and benchmark command before publication."]},"indexingState":"indexed","resultKey":"res_sAdWBwJDj-ssmqZDIShF1V_A","eligibilityReasons":[],"artifactBacked":true,"downloads":[]}