[FEAT] MatMul API #871
Merged
Merged
Conversation
yshekel
reviewed
Apr 22, 2025
yshekel
reviewed
Apr 22, 2025
yshekel
left a comment
yshekel
left a comment
Contributor
There was a problem hiding this comment.
Structure looks good for generalized mat-mul
I think not what we need in labrador (its Rq elements there in NTT domain) but we can start from that, @LeonHibnik ?
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
suggested changes
May 20, 2025
yshekel
left a comment
yshekel
left a comment
Contributor
There was a problem hiding this comment.
Good work, I left a few comments.
yshekel
reviewed
May 20, 2025
yshekel
reviewed
May 20, 2025
yshekel
reviewed
Jun 10, 2025
yshekel
reviewed
Jun 10, 2025
|
|
||
| // Process each matrix A in the batch with the same B matrix | ||
| for (size_t i = 0; i < batch_size; i++) { | ||
| ICICLE_CHECK(matrix_mult( |
Contributor
There was a problem hiding this comment.
Why do you need a batch in this test? The API doesn't support it (and it is not required)
yshekel
reviewed
Jun 10, 2025
yshekel
reviewed
Jun 10, 2025
yshekel
left a comment
yshekel
left a comment
Contributor
There was a problem hiding this comment.
Looks good, great tests but two problems:
(1) missing cpu implementation
(2) missing tests and code for PolyRing type
LeonHibnik
reviewed
Jun 11, 2025
LeonHibnik
left a comment
LeonHibnik
left a comment
Contributor
There was a problem hiding this comment.
Looks good overall, left some comments.
I think we should move the matmul and transpose implementations from vecops to matops, open to a discussion
| // Naive algorithm using taskflow | ||
|
|
||
| // Divide the problem among workers | ||
| const int nof_workers = 10; // get_nof_workers(config); |
Contributor
There was a problem hiding this comment.
Why is the nof_workers hardcoded?
| for (uint32_t j = 0; j < nof_cols_b; j++) { | ||
| // Initialize result element | ||
| uint64_t a_idx = config.columns_batch ? (i * nof_cols_a + 0) * stride : i * nof_cols_a + 0; | ||
| uint64_t b_idx = config.columns_batch ? (0 * nof_cols_b + j) * stride : 0 * nof_cols_b + j; |
Contributor
There was a problem hiding this comment.
Why is nof_cols_b here?
Contributor
Author
There was a problem hiding this comment.
legibility, to keep it analogous to the loop body below
| for (uint32_t i = row_start; i < row_end; i++) { | ||
| for (uint32_t j = 0; j < nof_cols_b; j++) { | ||
| // Initialize result element | ||
| uint64_t a_idx = config.columns_batch ? (i * nof_cols_a + 0) * stride : i * nof_cols_a + 0; |
Contributor
There was a problem hiding this comment.
Remove the + 0 and add a comment explaining it
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
reviewed
Jun 16, 2025
yshekel
approved these changes
Jun 16, 2025
LeonHibnik
approved these changes
Jun 16, 2025
LeonHibnik
pushed a commit
that referenced
this pull request
Jun 30, 2025
Co-authored-by: LNM <lisa@ingonyama.com> Co-authored-by: yshekel <yshekel@gmail.com>
|
Nice |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
cuda-backend-branch: lisa/matmul