MLPerf v6.1 Cisco 4x Xeon 6788P: llama3.1-8b INT8 Offline aggregate tok/s

MLPerf Inference v6.1 closed division, CPU-only (summary.csv accelerator_model_name is N/A). Values copied from summary.csv, not from a per-system mlperf_log_summary.txt. AGGREGATE throughput for the whole node, NOT per-request decode and NOT per-socket. Prompt length, output length, and batch size are not in the summary row. weight_data_types INT8. Framework column is PyTorch. Scenario: Offline. System: Cisco X410M8, 1-node-4S-GNR_86C. Four sockets, 86 cores each, 8 channels each. 2068.63 tok/s Offline is only about 1.68x the 2S Quanta 6788P row (1170.51), not 2x. Aggregate across the whole node, not per socket. measured_by vendor: these are submitter results inside the MLCommons closed division (Intel, Dell, Supermicro, Quanta, Red Hat, Cisco), not an independent re-run.

Four sockets, 86 cores each, 8 channels each. 2068.63 tok/s Offline is only about 1.68x the 2S Quanta 6788P row (1170.51), not 2x. Aggregate across the whole node, not per socket.