llama.cpp
CUDA/HIP: refractor mmqv to unify the calculation of nwarps and rows per block between host and device code.
#12177
Merged
Merged
CUDA/HIP: refractor mmqv to unify the calculation of nwarps and rows per block between host and device code. #12177
IMbackK
merged
5 commits into
ggml-org:master
from
IMbackK:refactor_mmqv
refractor mmqv to unify the calculation of nwarps and rows per block …
888ffc87
make cuda happy, as it dosent support calling host constexpr function…
a55d765d
Fix nits
b85a723a
Fix spelling of parameter
15f4dcaf
JohannesGaessler
approved these changes
on 2025-03-07
Update ggml/src/ggml-cuda/mmvq.cu
1b3894eb
Login to write a write a comment.
Login via GitHub