MLPerf v3.0: Dell r7515, 4x Qualcomm Cloud AI 100 Pro, BERT-99 Server 2724.39 samples/s
2724.39 completed samples/s, 4-card node, Server scenario under latency SLO
Fields
id
benchmarks/qaic-ai100-4x-mlperf-v3-0-bert99-server-samplesMetric
tps_aggregate
Value
2,724.39
Unit
completed samples/s, 4-card node, Server scenario under latency SLO
Model
bert-large-squad (bert-99)
Methodology
MLPerf Inference v3.0 CLOSED submission, Dell r7515 with 4x Qualcomm Cloud AI 100 Pro cards (q4_pro config, Qualcomm Apps SDK v1.8.3.7). BERT-large SQuAD INT8, Server scenario: value is 'completed samples/s' at the scheduled QPS that met latency SLO, read verbatim from performance/run_1/mlperf_log_summary.txt. Non-LLM encoder workload; see the matching offline record for why this exists (base AI100 has no public LLM benchmark).
Measured by
vendor
Reproducible
yes