Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings

Commit bb115d2

Browse filesBrowse files
authored
musa: override warp_size of musa device to 32 (ggml-org#12445)
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
1 parent 29fff30 commit bb115d2
Copy full SHA for bb115d2

File tree

Expand file treeCollapse file tree

1 file changed

+2
-0
lines changed
Filter options
Expand file treeCollapse file tree

1 file changed

+2
-0
lines changed

‎ggml/src/ggml-cuda/ggml-cuda.cu

Copy file name to clipboardExpand all lines: ggml/src/ggml-cuda/ggml-cuda.cu
+2Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -262,6 +262,8 @@ static ggml_cuda_device_info ggml_cuda_init() {
262262
id, prop.name, prop.gcnArchName, info.devices[id].cc & 0xffff,
263263
device_vmm ? "yes" : "no", prop.warpSize);
264264
#elif defined(GGML_USE_MUSA)
265+
// FIXME: Ensure compatibility with varying warp sizes across different MUSA archs.
266+
info.devices[id].warp_size = 32;
265267
// TODO: refine the .cc to reflect MUSA's actual CC capabilities
266268
info.devices[id].smpbo = prop.sharedMemPerBlockOptin;
267269
info.devices[id].cc = 100*prop.major + 10*prop.minor;

0 commit comments

Comments
0 (0)
Morty Proxy This is a proxified and sanitized view of the page, visit original site.