MLPerf Inference v5.1 open results — NVIDIA Jetson AGX Thor llama3.1-8b-edge SingleStream (raw log)

Fields

id

sources/mlperf-v5-1-nvidia-thor-llama31-8b-edge-singlestream

URL

https://github.com/mlcommons/inference_results_v5.1/blob/main/open/NVIDIA/results/Jetson-Thor_TRT/llama3.1-8b-edge/SingleStream/performance/run_1/mlperf_log_summary.txt

Publisher

MLCommons / NVIDIA

Kind

benchmark

Published

2025-09-09

Accessed

2026-10-17

notes

Open-division submission. 1x Jetson AGX Thor Developer Kit 128GB, TensorRT 10.8, CUDA 12.8, FP4 weights. SingleStream only (no Offline row in v5.1). Log prints 90th-percentile first-token latency and time-to-output-token separately, both VALID. Quoted verbatim 2026-10-17.

Notes

Open-division submission. 1x Jetson AGX Thor Developer Kit 128GB, TensorRT 10.8, CUDA 12.8, FP4 weights. SingleStream only (no Offline row in v5.1). Log prints 90th-percentile first-token latency and time-to-output-token separately, both VALID. Quoted verbatim 2026-10-17.

Cross-references

This record references no other records.