2 articles · Updated 2026-10-01

Tuning

Power limits, multi-GPU splits, cooling. Small changes that get more out of the hardware you already have.

Two different GPUs, one model

2026-10-01

How llama.cpp splits a model across mismatched cards, and what it means for speed.

Power-limit your card, keep your speed

2026-10-01

Generation speed depends on memory bandwidth far more than on GPU power. A lower power limit often costs very little.