DisableUnrollLoops should ignore DisableLLVMLoopOpt for PTX#5900
Closed
steven-johnson wants to merge 25 commits intomasterfrom
Closed
DisableUnrollLoops should ignore DisableLLVMLoopOpt for PTX#5900steven-johnson wants to merge 25 commits intomasterfrom
steven-johnson wants to merge 25 commits intomasterfrom
Conversation
It's not entirely clear why, but setting the `DisableUnrollLoops` flag to `true` in the PTX backend can produce Cuda code that will fail at runtime, at least for certain drivers (e.g.: some of the tests in correctness_gpu_dynamic_shared when running under NVidia v460.x).
This was referenced Apr 14, 2021
Closed
Member
|
This doesn't seem like the right fix. The code with loop opts disabled is much simpler and clearer, yet it crashes. You said it was driver-dependent, right? Maybe this is a bug inside the cuda driver? |
Member
|
To be clearer, I think the reason this isn't the right fix is that I would like disable_llvm_loop_opt to work for the bodies of cuda kernels, because I think a lot of the time that generates better kernels. |
Contributor
Author
Yeah, probably correct -- but this is in the standard nvidia (closed-source) linux driver v460 (which Ubuntu 20 installs by default), and it's been out for a while with no updates. Maybe we should file a bug? |
steven-johnson
added
the
skip_buildbots
steven-johnson
removed
the
skip_buildbots
steven-johnson
added
the
skip_buildbots
Contributor
Author
|
Definitely not the right fix, closing |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
It's not entirely clear why, but setting the
DisableUnrollLoopsflag totruein the PTX backend can produce Cuda code that will fail at runtime, at least for certain drivers (e.g.: some of the tests in correctness_gpu_dynamic_shared when running under NVidia v460.x).