Skip to content

Conversation

@slaren
Copy link
Member

@slaren slaren commented Nov 2, 2024

The goal is to improve readability of the markdown output by skipping repeated column values. Not sure if it is actually better.

For example:

./llama-bench -fa 0,1 -p 128,512 -m models/7B/ggml-model-f16.gguf -m models/7B/ggml-model-Q4_0.gguf
model size params backend ngl fa test t/s
llama 7B F16 12.55 GiB 6.74 B CUDA 99 0 pp128 4258.54 ± 11.35
pp512 5613.05 ± 15.36
tg128 60.14 ± 0.04
1 pp128 4376.25 ± 7.00
pp512 6078.50 ± 11.93
tg128 61.07 ± 0.35
llama 7B Q4_0 3.56 GiB 0 pp128 4925.08 ± 13.67
pp512 5894.38 ± 5.63
tg128 161.10 ± 0.30
1 pp128 5076.93 ± 6.05
pp512 6422.28 ± 5.30
tg128 167.63 ± 0.55

@slaren slaren closed this Jul 3, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants