[Library] Add cublas library #404
Merged
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
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 19, 2024
The issue is caused by a wrong layout for the bias tensor. For example, we consider a bias tensor of shape (64, ) and its layout can be written as `(64, ): (1, )` However, we can expand the layout by adding axes with 1-shape. For example, `(64, 1):(1, 1)` Since the shape is equal to 1, the stride can be any number. The stride corresponding to the 1-shape actually doesn't affect the computation of the address. But two strides that are equal to one will influence the instruction selection, and the invalid memory instruction leads to the misaligned access. To fix this issue, we force the stride paired with 1-shape to be 0. The layout is equivalent when computing the memory address, and this will help the compiler make the right decision in the instruction selection pass. closes #404 Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 20, 2024
The issue is caused by a wrong layout for the bias tensor. For example, we consider a bias tensor of shape (64, ) and its layout can be written as `(64, ): (1, )` However, we can expand the layout by adding axes with 1-shape. For example, `(64, 1):(1, 1)` Since the shape is equal to 1, the stride can be any number. The stride corresponding to the 1-shape actually doesn't affect the computation of the address. But two strides that are equal to one will influence the instruction selection, and the invalid memory instruction leads to the misaligned access. To fix this issue, we force the stride paired with 1-shape to be 0. The layout is equivalent when computing the memory address, and this will help the compiler make the right decision in the instruction selection pass. closes #404 Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 26, 2024
The issue is caused by a wrong layout for the bias tensor. For example, we consider a bias tensor of shape (64, ) and its layout can be written as `(64, ): (1, )` However, we can expand the layout by adding axes with 1-shape. For example, `(64, 1):(1, 1)` Since the shape is equal to 1, the stride can be any number. The stride corresponding to the 1-shape actually doesn't affect the computation of the address. But two strides that are equal to one will influence the instruction selection, and the invalid memory instruction leads to the misaligned access. To fix this issue, we force the stride paired with 1-shape to be 0. The layout is equivalent when computing the memory address, and this will help the compiler make the right decision in the instruction selection pass. closes #404 Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
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 add cublas to hidet.
Check the
tests/cuda/test_cublas.pyfor the usage of cublas in hidet.