verifiedhigh · 0.85updated 2026-10-14

MLPerf v5.1 IBM Cloud 1x L40S vLLM: llama3.1-8b Server completed tokens/s

1207.14 tok/s (single L40S, Server scenario)

Fields

benchmarks/mlperf-v5-1-ibm-1xl40s-vllm-llama31-8b-server-tokens

Metric

tps_aggregate

Value

1,207.14

Unit

tok/s (single L40S, Server scenario)

Model

llama3.1-8b

Methodology

MLPerf Inference closed-division submission. v5.1, IBM Cloud 1x L40S, vLLM 0.10.0. Server: 'Completed tokens per second: 1207.14', 9.37 scheduled samples/s. Values quoted verbatim from the submission's mlperf_log_summary.txt (performance/run_1). AGGREGATE throughput across all accelerators in the system, NOT per-request.

Measured by

vendor

Reproducible

yes

Cross-references

MLPerf v5.1 IBM Cloud 1x L40S vLLM: llama3.1-8b Server completed tokens/s

Fields

Cross-references

Accelerators (1)

Engine (1)

Interconnect (1)