SOURCE-LINKED INTELLIGENCE

llama3.1-8b — Server — H200

MLCommons MLPerf · observation · Publication date unknown

7376.5976662324 Tokens/s; MLPerf v6.1, closed.

Read original source ↗ Open in workspace

recordType

benchmark

region

Global

value

7376.5976662324

unit

Tokens/s

dimensions

{"division":"closed","location":"./closed/Naeem Khoshnevis/results/H200/llama3.1-8b/Server/performance/run_1","metric":"Performance_Result","model":"llama3.1-8b","precision":"fp8 (static per-tensor PTQ, NVIDIA ModelOpt)","release":"v6.1","scenario":"Server","submissionId":null,"submitter":"Naeem Khoshnevis","suite":"datacenter","system":"H200","unit":"Tokens/s"}