Argonne A100-SXM-40GB x4 vLLM meta-llama/Llama-2-70b-hf b1
59.74 tok/s aggregate (batch*(in+out)/latency)
Fields
id
benchmarks/argonne-a100sxm40-vllm-llama2-70b-fp16-4x-b1-128io-tpsMetric
tps_aggregate
Value
59.74
Unit
tok/s aggregate (batch*(in+out)/latency)
Model
meta-llama/Llama-2-70b-hf
Methodology
THIRD-PARTY MEASURED (Argonne LLM-Inference-Bench, arXiv:2411.00136, All_results.csv). vLLM on NVIDIA A100 SXM 40GB x4, FP16/BF16, 128 in / 128 out, batch 1. Throughput = batch*(in+out)/latency = AGGREGATE across batch. Latency 4.29s.
Measured by
third_party
Reproducible
yes
Thermal state
unknown
notes
Argonne hardware matrix: A100 SXM 40GB x4/node. Framework=vLLM.
Notes
Argonne hardware matrix: A100 SXM 40GB x4/node. Framework=vLLM.