Benchmark archive
Ternary Bonsai 2 27B PQ2_0
RTX 2080 Ti · Vulkan · 1 runs
| Run (UTC) | geist | llama.cpp | pp512 tok/s | vs llama | tg64 tok/s | vs llama | Files |
|---|---|---|---|---|---|---|---|
| 2026-10-12 01:26 | 4d6e2696a | adfffbe41 | 635.5 | +16 % | 40.0 | +36 % | 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.