Argonne MI250 vLLM Mistral-7B b64
3903.4 tok/s aggregate
Fields
id
benchmarks/argonne-mi250-vllm-mistral-7b-fp16-b64-128io-tpsMetric
tps_aggregate
Value
3,903.4
Unit
tok/s aggregate
Model
mistralai/Mistral-7B-v0.1
Methodology
THIRD-PARTY MEASURED (Argonne LLM-Inference-Bench arXiv:2411.00136). vLLM, FP16/BF16, 128in/128out. throughput=batch*(in+out)/e2e latency = AGGREGATE over batch. 1 tile bs64. CSV 'AMD MI250 GPU' = ALCF Polaris MI250 (2-tile GCD card, 128GB HBM2e). N= number of tiles/cards used.
Measured by
third_party
Reproducible
yes
notes
CORRECTED 2026-10-15: value re-read from All_results.csv
Notes
CORRECTED 2026-10-15: value re-read from All_results.csv