A user reports running Gemma 4 31B Q6 on two NVIDIA 9060 XT 16GB cards, achieving consistent throughput of 8-9 tokens per second. They note the performance is usable but below expectations, suggesting potential optimizations or hardware limitations.
Gemma 4 31B Q6 Runs at 8-9 t/s on Dual 9060 XT Cards
from English