llama.cpp CUDA P106-100 llama-2-7b Q4_0 tg128 FA
30.64 tok/s (tg128)
Fields
id
benchmarks/llamacpp-scoreboard-p106-100-fa-tg128Metric
decode_tok_s
Value
30.64
Unit
tok/s (tg128)
Model
llama-2-7b.Q4_0.gguf (ggml-org, 3.56 GiB, 6.74B params)
Methodology
COMMUNITY-SCOREBOARD (llama.cpp discussion #15013, maintainer-curated CUDA 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, CUDA. Per-request batch-1, not an aggregate across concurrent requests. Flash attention on (-fa 1). Build 5fd160b, reporter @pebaryan. Printed mean 30.64. Re-read of the discussion page on 2026-10-17.
Measured by
third_party
Reproducible
yes
notes
P106-100, 6 GB / GDDR5 / 192 bit, build 5fd160b, reporter @pebaryan. token generation cell.
Notes
P106-100, 6 GB / GDDR5 / 192 bit, build 5fd160b, reporter @pebaryan. token generation cell.