MLPerf v6.1 Quanta 1-node-4x-BMG-B70: llama3.1-8b Offline 4977.67 tok/s
4977.67 tok/s aggregate Offline
Fields
id
benchmarks/mlperf-v6-1-quanta-4xarcb70-llama31-8b-offline-tokensMetric
tps_aggregate
Value
4,977.67
Unit
tok/s aggregate Offline
Model
Llama-3.1-8B
Methodology
MLPerf Inference v6.1 published result (Quanta 1-node-4x-BMG-B70, 4x Arc Pro B70; software: PyTorch). System-AGGREGATE output tokens/s for the named scenario, submitter-measured under MLCommons audit. Highest of the three v6.1 4xB70 8B Offline submissions (Intel 4613.28, Supermicro 4596.62).
Measured by
vendor
Reproducible
yes
notes
ENGINE_ID NOTE: pinned to 'vllm' as the closest existing engine record; the submission's actual stack is named in this methodology string.
Notes
ENGINE_ID NOTE: pinned to 'vllm' as the closest existing engine record; the submission's actual stack is named in this methodology string.