llama.cpp ROCm Instinct MI25 llama-2-7b Q4_0 pp512 noFA
409.83 tok/s (pp512)
Fields
id
benchmarks/llamacpp-scoreboard-instinct-mi25-nofa-pp512Metric
prefill_tok_s
Value
409.83
Unit
tok/s (pp512)
Model
llama-2-7b.Q4_0.gguf (ggml-org, 3.56 GiB, 6.74B params)
Methodology
COMMUNITY-SCOREBOARD (llama.cpp discussion #15021, maintainer-curated ROCm backend). llama-bench llama-2-7b.Q4_0.gguf (ggml-org, 3.56 GiB, 6.74B params), pp512 and tg128, n_gpu_layers 99, batch 1, full GPU offload, ROCm. Per-request batch-1, not an aggregate across concurrent requests. Flash attention off (-fa 0). Build 2739a71, reporter @8XXD8. Printed mean 409.83. Re-read of the discussion page on 2026-10-17.
Measured by
third_party
Reproducible
yes
notes
Instinct MI25, 16 GB / HBM2 / 2048 bit, build 2739a71, reporter @8XXD8. prompt processing cell.
Notes
Instinct MI25, 16 GB / HBM2 / 2048 bit, build 2739a71, reporter @8XXD8. prompt processing cell.