optimize com.microsoft.MatMulNbits operator #28504
Open
+75
−107
Conversation
This file contains 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
### Details:
- use convert instead of convert_like op, it help disabled const
folding and run online int2/4/8 dequantize rather than const
folding as complie time, benefits compile memory usage and
inference latency.
- use zero point as uint2/4/8, it trigled oneDNN kernel, much
benefits the GPU performance.
|
build_jenkins |
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Outdated suggestions cannot be applied.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.
This PR is doing some optimization work on onnxfrontend com.microsoft.MatMulNbits operators
with this changes:
we tested this changes along with another PR #28163 , and confirmed phi3/llama3 INT4 model run well in LNL.