llama3.1

on RTX 3090

136.2

tok/sdecode

RTX 3090 (24GB) + AMD EPYC 7663 56-Core Processor (56c) + 252GB

rank in tier

5/14RTX 3090 runstop 36%

best workload

chat-short

where the rig flew

slowest workload

agent-trace127.5 tok/s

where the rig struggled

backend

ollamaollama is llama.cpp under the hood; pure llama.cpp tends to nudge a few % faster.

faster than

r1 on RTX 4090133.8 tok/s
Qwen2.5-14B-Instruct on RTX 5090133.3 tok/s
Llama-3.1-8B-Instruct-4bit on M3 Ultra130.2 tok/s

This is a shareable view of one signed llm-speed suite-suite-v1 run. Numbers are not edited. The raw record has every workload, the public key, and the fingerprint hash.