[mlir][Target][NVPTX] Add fatbin support to NVPTX compilation. #65398
Conversation
This file contains hidden or 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
Currently, the NVPTX tool compilation path only calls `ptxas`; thus, the GPU running the binary must be an exact match of the arch of the target, or else the runtime throws an error due to the arch mismatch. This patch adds a call to `fatbinary`, creating a fat binary with the cubin object and the PTX code, allowing the driver to JIT the PTX at runtime if there's an arch mismatch.
|
Is this something that should be under the control of the target attribute with an option there? The user may want to include the PTX or not... |
joker-eph
reviewed
Sep 5, 2023
| @@ -184,7 +184,7 @@ class NVPTXSerializer : public SerializeGPUModuleBase { | |||
| // 1. The toolkit path in `targetOptions`. | |||
| // 2. In the system PATH. | |||
| // 3. The path from `getCUDAToolkitPath()`. | |||
| std::optional<std::string> findPtxas() const; | |||
| std::optional<std::string> findTool(StringRef tool); | |||
There was a problem hiding this comment.
Please document the tool arg
I'm not sure if it should be in the target attribute, as we tried to remove those kinds of flags from it. However, I'm thinking something more like a compilation option passed through |
joker-eph
approved these changes
Sep 7, 2023
avillega
pushed a commit
to avillega/llvm-project
that referenced
this pull request
Sep 11, 2023
…65398) Currently, the NVPTX tool compilation path only calls `ptxas`; thus, the GPU running the binary must be an exact match of the arch of the target, or else the runtime throws an error due to the arch mismatch. This patch adds a call to `fatbinary`, creating a fat binary with the cubin object and the PTX code, allowing the driver to JIT the PTX at runtime if there's an arch mismatch.
Guzhu-AMD
pushed a commit
to GPUOpen-Drivers/llvm-project
that referenced
this pull request
Sep 14, 2023
Local branch amd-gfx 0d8d006 Fix build warning in SIInsertWaterfall Remote branch main c16adb0 [mlir][Target][NVPTX] Add fatbin support to NVPTX compilation. (llvm#65398)
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.
Currently, the NVPTX tool compilation path only calls
ptxas; thus, the GPU running the binary must be an exact match of the arch of the target, or else the runtime throws an error due to the arch mismatch.This patch adds a call to
fatbinary, creating a fat binary with the cubin object and the PTX code, allowing the driver to JIT the PTX at runtime if there's an arch mismatch.This patch is needed to start migrating the Integration Tests, otherwise there will be a runtime error due to architecture mismatch.