fix nvrtc usage #60943
Closed
fix nvrtc usage #60943
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! |
risemeup1
reviewed
Jan 18, 2024
| @@ -335,7 +335,7 @@ bool GPUDeviceCode::Compile(bool include_path) { | |||
| DeviceContextPool::Instance().Get(place_)); | |||
| int compute_capability = dev_ctx->GetComputeCapability(); | |||
| std::string compute_flag = | |||
| "--gpu-architecture=compute_" + std::to_string(compute_capability); | |||
| "--gpu-architecture=sm_" + std::to_string(compute_capability); | |||
| std::vector<const char*> options = {"--std=c++11", compute_flag.c_str()}; | |||
There was a problem hiding this comment.
-gpu-architecture=sm_XX：这个选项会让编译器直接生成目标GPU架构的机器码（SASS）。使用sm_XX可以避免运行时的JIT编译步骤，但是这样生成的代码只能在相同架构的GPU上运行。使用compute_XX可以使得生成的代码在不同架构的GPU上都能运行。。有考虑过这个问题吗？
|
Sorry to inform you that f14857f's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually. |
不過，我再深入調研了代碼，發現更本的問題不在 NVRTC 的使用 |
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Outdated suggestions cannot be applied.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
Bug fixes
PR changes
Others
Description
The phi uses nvrtc to implement JIT-compilation. However, it uses
--gpu-architecture=cmopute_rather than--gpu-architecture=sm_. The difference is that compute_ only generates PTX code instead of machine code (ie, SASS). The sm_ has both.So, the workflow is nvrtc generates PTX code, then invokes the driver to generate machine code.
The problem is if the driver version doesn't match CUDA version, the driver cannot recognize the format of PTX code, so it triggers “provided ptx was compiled with an unsupported toolchain” error. This issue cannot be resolved by “CUDA forward compatibility enabled”.
The only solutions are two:
Many peoples cannot upgrade the driver frequently, so this patch is the 2nd solution.