RTX PRO 2000 Blackwell Ollama: llama3.1:8b eval rate
59.05 tok/s (eval rate, per-request, batch 1)
Fields
id
benchmarks/dm-pro2000-ollama-llama31-8b-evalMetric
decode_tok_s
Value
59.05
Unit
tok/s (eval rate, per-request, batch 1)
Model
llama-3-1-8b
Methodology
Ollama on one RTX PRO 2000 Blackwell 16 GB. Ollama version not printed on this page (the sibling PRO 6000 page is 0.13.5; this one is not asserted to match). Quantization is the default Ollama pull, labelled INT4, exact GGUF quant not printed, format_id null. Prompt is the single question 'What is GPU?'; output length is whatever the model produced. Eval rate is per-request decode, batch size 1, NOT aggregate.
Measured by
third_party
Reproducible
yes
notes
4.9 GB, GPU util 88%.
Notes
4.9 GB, GPU util 88%.