CANN: supports out_prod operator for F32 and F16 #17406
Open
+95
−0
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
TianHao324
changed the title
Author
|
test result |
noemotiovon
reviewed
Nov 20, 2025
Collaborator
noemotiovon
left a comment
noemotiovon
left a comment
There was a problem hiding this comment.
LGTM, just a minor issue.
ggml/src/ggml-cann/aclnn_ops.cpp
Outdated
|
|
||
| const int64_t i12 = i2; | ||
| const int64_t i13 = i3; | ||
| aclTensor *accumulator = ggml_cann_create_tensor( |
Collaborator
There was a problem hiding this comment.
The result of ggml_cann_create_tensor should be acl_tensor_ptr, not aclTensor*.
ggml/src/ggml-cann/aclnn_ops.h
Outdated
| */ | ||
| void ggml_cann_out_prod(ggml_backend_cann_context & ctx, ggml_tensor * dst); | ||
|
|
||
| void ggml_cann_out_prod_fp(ggml_backend_cann_context & ctx, ggml_tensor * dst); |
ggml/src/ggml-cann/aclnn_ops.cpp
Outdated
| #include <aclnnop/aclnn_index_select.h> | ||
| #include <aclnnop/aclnn_clamp.h> | ||
| #include <aclnnop/aclnn_threshold.h> | ||
| #include <aclnnop/aclnn_ger.h> |
Collaborator
There was a problem hiding this comment.
You should use
find ggml/src/ggml-cann -iname ".cpp" -o -iname ".h" | xargs clang-format -i
to format code.
ggml/src/ggml-cann/aclnn_ops.cpp
Outdated
| #include <aclnnop/aclnn_index_select.h> | ||
| #include <aclnnop/aclnn_clamp.h> | ||
| #include <aclnnop/aclnn_threshold.h> | ||
| #include <aclnnop/aclnn_ger.h> |
Collaborator
There was a problem hiding this comment.
You should use
find ggml/src/ggml-cann -iname ".cpp" -o -iname ".h" | xargs clang-format -i
to format code.
ggml/src/ggml-cann/aclnn_ops.cpp
Outdated
| dst->nb, | ||
| 2); | ||
|
|
||
| GGML_CANN_CALL_ACLNN_OP(ctx, InplaceZero, accumulator); |
Collaborator
There was a problem hiding this comment.
Currently, InplaceZero is being called on each iteration of the for loop. I believe we can just call it once on dst before the loop.
github-actions
bot
added
the
ggml
TianHao324
force-pushed
the
out_prod
branch
3 times, most recently
from
815e770 to
5d9578a
Compare
Collaborator
|
Thank you for your contribution! :) |
* DGX Spark: UMA support * Updates from PR feedback * More PR feedback cleanup * Update ggml/src/ggml-cuda/ggml-cuda.cu Co-authored-by: Georgi Gerganov <[email protected]> * Remove trailing whitespace * Update ggml/src/ggml-cuda/ggml-cuda.cu --------- Co-authored-by: Georgi Gerganov <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
The CANN backend supports floating-point product calculations.