Benchmark archive
Qwen3.8 27B Q4_0
AMD Ryzen 9 9950X · CPU · 1 runs
| Run (UTC) | geist | llama.cpp | pp512 tok/s | vs llama | tg64 tok/s | vs llama | Files |
|---|---|---|---|---|---|---|---|
| 2026-10-09 17:18 | bce40a4bd | d81235049 | 38.9 | -28 % | 4.8 | +8 % | report · raw |
Every run of this cell, newest first. pp512: prefill of 512 tokens; tg64: decode of 64 tokens at depth 512; median of 12 samples. The raw file holds every sample and the full host, engine and model provenance. Back to the matrix.