Argonne LLM-Inference-Bench: MI250 x1, vLLM 0.6.1, Llama-2-7B FP16, 128in/128out, batch 64
4477.1 tok/s aggregate across 64 batched requests
Fields
id
benchmarks/argonne-mi250-vllm-llama2-7b-fp16-b64-128io-tpsMetric
tps_aggregate
Value
4,477.1
Unit
tok/s aggregate across 64 batched requests
Model
meta-llama/Llama-2-7b-hf
Methodology
THIRD-PARTY MEASURED (Argonne LLM-Inference-Bench, arXiv:2411.00136, All_results.csv). vLLM benchmark_throughput.py, FP16/BF16, 128 in / 128 out. Throughput = batch*(in+out)/end-to-end latency = AGGREGATE over the batch, NOT per-request decode. Batch 64, 1x MI250. Latency 3.660s.
Measured by
third_party
Reproducible
yes
notes
Peak point on the paper's Fig.35 sweep for Llama-2-7B; batch-64 aggregate.
Notes
Peak point on the paper's Fig.35 sweep for Llama-2-7B; batch-64 aggregate.