GPU Hadamard for large N #1879
Merged
GPU Hadamard for large N #1879
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
|
Looks great! As per offline discussion let's move this into the primitive instead. |
angeloskath
force-pushed
the
big-gpu-hadamard
branch
from
d61c204 to
90f66b6
Compare
Member
|
This should be fine to review and merge now. The kernel is a bit faster than copying and calling the contiguous one and it also has the benefit of being completely in-place which the transpose copy can't do.
The 16K elements in-place transform was not launching correctly (max threads per thread group 832) so I reduced the limit to 8K. @barronalex if you remember encountering the same issue before let me know how you did fix it 🤔. Edit: The plot is in GB/s not MB/s |
awni
reviewed
May 1, 2025
angeloskath
force-pushed
the
big-gpu-hadamard
branch
2 times, most recently
from
fb6f761 to
61ffdf0
Compare
angeloskath
added a commit
that referenced
this pull request
May 2, 2025
angeloskath
force-pushed
the
big-gpu-hadamard
branch
from
61ffdf0 to
00bad20
Compare
angeloskath
force-pushed
the
big-gpu-hadamard
branch
from
00bad20 to
4813494
Compare
faisalmemon
pushed a commit
to faisalmemon/mlx
that referenced
this pull request
Oct 30, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add support for large Hadamard transforms on the GPU.
For$N=2^{24}$ the GPU version is about 50x faster than the CPU: