Skip to content

HIP : route Q6_K through MMQ up to ne11 = 1024 on RDNA 3.5 #38

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Nyovelt wants to merge 1 commit into
base: master
Choose a base branch
from
Open

HIP : route Q6_K through MMQ up to ne11 = 1024 on RDNA 3.5 #38

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions ggml/src/ggml-cuda/mmq.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -553,6 +553,11 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11, int64_t
case GGML_TYPE_Q2_K:
return ne11 <= 128;
case GGML_TYPE_Q6_K:
// RDNA 3.5 (gfx1151): dequantize + hipBLASLt loses to MMQ up to ne11 = 1024; at 2048 the two
// paths trade places per shape (K = 4096 favours hipBLASLt, K >= 12288 favours MMQ), see PR.
Comment on lines +556 to +557
if (GGML_CUDA_CC_IS_RDNA3_5(cc)) {
return ne11 <= 1024;
}
return ne11 <= (GGML_CUDA_CC_IS_RDNA3_0(cc) ? 128 : 256);
case GGML_TYPE_IQ2_XS:
case GGML_TYPE_IQ2_S:
Expand Down