-
Notifications
You must be signed in to change notification settings - Fork 271
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
【Hackathon4 No.35】为 Paddle 优化 prelu op 在 GPU 上的计算性能 #370
Merged
Merged
Changes from all commits
Commits
Show all changes
3 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,93 @@ | ||
# Prelu OP性能优化设计文档 | ||
|
||
|
||
| 基本信息 | 内容 | | ||
| ------------------------------------------------------------ |--------------------------------------| | ||
| 提交作者<input type="checkbox" class="rowselector hidden"> | thunder95 | | ||
| 提交时间<input type="checkbox" class="rowselector hidden"> | 2023-02-22 | | ||
| 版本号 | V1.0 | | ||
| 依赖飞桨版本<input type="checkbox" class="rowselector hidden"> | PaddleDevelop | | ||
| 文件名 | 20230222_prelu_op_optimization.md<br> | | ||
|
||
|
||
# 1 背景与意义 | ||
|
||
目前Paddle中的Prelu算子仍旧通过内部循环方式实现,没有用到一些性能优化的技巧,存在性能优化的空间。 | ||
|
||
## 1.1 飞桨现状 | ||
|
||
当前性能如下表(基于PaddlePaddle develop分支): | ||
|
||
目前的实现有一定的性能优化空间,可以加入一些性能优化的技巧。当前forward性能如下表: | ||
|
||
| Case No. | device | input_shape | input_type | weight_type | Paddle Perf(ms) | | ||
|---|---|---|---|---|---| | ||
| 1 | RTX 2070s | [8L, 1024L, 3072L] | float32 | [1L] | 0.8584 | | ||
| 2 | RTX 2070s | [8L, 1024L, 3072L] | float32 | [1024L] | 1.1135 | | ||
| 3 | RTX 2070s | [8L, 1024L, 3072L] | float16 | [1L] | 0.62442 | | ||
| 4 | RTX 2070s | [8L, 1024L, 3072L] | float16 | [1024L] | 0.87672 | | ||
|
||
API文档 https://www.paddlepaddle.org.cn/documentation/docs/zh/api/paddle/nn/PReLU_cn.html#prelu | ||
|
||
## 1.2 业内方案调研 | ||
|
||
Pytorch中对Prelu算子的实现基于GPU计算, forward整体性能如下(基于pytorch v1.12): | ||
|
||
| Case No. | device | input_shape | input_type | weight_type | Pytorch Perf(ms) | | ||
|---|---|---|---|---|---| | ||
| 1 | RTX 2070s | [8L, 1024L, 3072L] | float32 | [1L] | 0.64366 | | ||
| 2 | RTX 2070s | [8L, 1024L, 3072L] | float32 | [1024L] | 0.83144 | | ||
| 3 | RTX 2070s | [8L, 1024L, 3072L] | float16 | [1L] | 0.31887 | | ||
| 4 | RTX 2070s | [8L, 1024L, 3072L] | float16 | [1024L] | 0.84326 | | ||
|
||
## 1.3 对比分析 | ||
|
||
目前Paddle与Pytorch的API设计方案相似,两种case下测试pytorch性能更优,理论上可以通过线程配置,或向量化读取和写入等手段进行优化,进一步提升算子性能。 | ||
Pytorch通过grid和block的优化配置性能明显优于paddle,paddle目前还是1d内部循环的方式实现。 | ||
|
||
# 2 设计方案与性能预期 | ||
|
||
## 2.1 关键模块与性能提升点 | ||
|
||
通过使用飞桨内部的Elementwise Kernel来进行计算。通过向量化读取、向量化写入以及gpu_launch_config.h中的线程配置方法对算子进行优化, 使用cuda内置函数后预计比当前算子提升%20以上。 | ||
|
||
## 2.2 Host端计算流程 | ||
|
||
通过gpu_launch_config.h中的线程配置方法配置1D线程。 | ||
|
||
## 2.4 Device端计算流程 | ||
|
||
设备端通过kps::ReadData和kps::WriteData对数据进行读写,再对每个值进行prelu计算。 | ||
|
||
# 3 测试和验收的考量 | ||
|
||
参考:[算子性能优化验收标准](http://agroup.baidu.com/paddle-perf/md/article/4892913) | ||
|
||
|
||
|
||
# 4 可行性分析和排期规划 | ||
|
||
时间和开发排期规划,主要milestone | ||
|
||
| No. | 开发内容 | 预期时间 | | ||
|---|---|---| | ||
| 1 | 理清Paddle中OP设计思路,同类产品中最佳设计方案 | 2023-02-22 | | ||
| 2 | 完成开发文档设计 | 2023-02-22 | | ||
| 3 | prelu优化实现 | 2023-02-23 | | ||
| 3 | 完成代码开发工作,并通过线程CI测试 | 2023-02-24 | | ||
|
||
|
||
|
||
# 5 影响面 | ||
|
||
待优化的算子独立运行,不涉及其他算子和模块的修改,API设计与之前保持一致。 | ||
|
||
|
||
# 名词解释 | ||
|
||
|
||
# 附件及参考资料 | ||
|
||
[1]. [OP Benchmark使用指南](https://github.com/PaddlePaddle/benchmark/blob/master/api/README.md) | ||
|
||
|
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
请补充下FP16的性能测试数据,优化时要考虑FP16的性能情况。
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@JamesLim-sy 已补充