[CINN][Add Backend Pass Comment No.10] Add comment for replace_cross_thread_reduction #70227
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
|
你的PR提交成功，感谢你对开源项目的贡献! |
| /** | ||
| * A pass that optimizes cross-thread reduction operations on GPU by replacing them with more efficient implementations. | ||
| * | ||
| * [Detailed application scenario] |
There was a problem hiding this comment.
方括号的为模板的提示，在正式注释里删掉即可
Comment on lines
28
to
30
| * Replace cross thread reduction to external call. | ||
| */ | ||
| void ReplaceCrossThreadReduction(ir::LoweredFunc fn); |
Comment on lines
71
to
75
| * TODO: | ||
| * - Support more reduction operations (e.g., custom reduction functions) | ||
| * - Add dynamic selection of reduction methods based on input size | ||
| * - Optimize shared memory allocation for better bank conflicts avoidance | ||
| * - Add support for multi-warp reductions within a block |
There was a problem hiding this comment.
原函数没有的TODO不用加了，这个是可选项
|
Sorry to inform you that f48074a's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually. |
Comment on lines
35
to
40
| * This pass is applicable in scenarios where multiple GPU threads need to perform reduction operations (like sum, max, min) | ||
| * across thread boundaries. These scenarios are common in deep learning workloads, particularly in operations like: | ||
| * - Computing sum/mean across feature dimensions | ||
| * - Global pooling operations | ||
| * - Softmax normalization | ||
| * - Gradient aggregation in distributed training |
There was a problem hiding this comment.
这个描述得太高层了，需要描述一下在后端IR优化上的场景
| * | ||
| * | ||
| * When applied, this pass will: | ||
| * 1. Identify reduction operations in GPU-bound loops |
Comment on lines
73
to
77
| * for (i = 0; i < 1024; i++) { | ||
| * if (i < n) { | ||
| * sum += data[i]; | ||
| * } | ||
| * } |
There was a problem hiding this comment.
ir里面应该体现一下gpu thread bind
|
PR-CI-Codestyle-Check 需要通过 |
|
@KDZZZZZZ PR-CI-Codestyle-Check 还没有通过，请再更新下。
|
test=document_fix
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.
PR types
CINN
PR changes
Others
Description
为replace_cross_thread_reduction Pass添加了注释