Fix abort with HIP backend for ROCm 5.0.2 and beyond #4873
Merged
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
Rombur
approved these changes
Mar 15, 2022
Comment on lines
+60
to
+61
| const char empty[] = ""; | ||
| __assert_fail(msg, empty, 0, empty); |
There was a problem hiding this comment.
Shouldn't this be guarded by a check for HIP_VERSION since we only require 4.5.0?
There was a problem hiding this comment.
The function exists in 4.5.0 but only starting with 5.0.2 the message is printed.
There was a problem hiding this comment.
HIP version was the same for ROCm 5.0.0 and 5.0.2 on Crusher
cz4rs
approved these changes
Mar 21, 2022
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Outdated suggestions cannot be applied.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.
The error message provided to
Kokkos::abortwas not printed when called on the device-side.The changes proposed (using
__assert_failinstead ofprintffollowed byabort) seem to do the charm with ROCm 5.0.2Unfortunately there is no "native" macro we can use to guard the workaround in the unit tests so I just added a comment. We could in principle do that at configuration time but I was a bit hesitant because it is hacky and we haven't done it for any of the toolchains yet.