SIGN IN SIGN UP

ggml-cuda: tune RDNA3 Q6_K MMVQ nwarps (llama/23349)

R
ravel7524 committed
0a0a34287e84641c34679c1a38fed299a56e9d4b
Parent: 37f1720
Committed by Georgi Gerganov <ggerganov@gmail.com> on 5/25/2026, 9:26:07 AM