-
Notifications
You must be signed in to change notification settings - Fork 14.3k
[flang][NFC] Restrict -funroll-loops tests to known working targets #123939
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
Conversation
@mustartt did you intend the issue to be for the failing tests or that the flag does not produce the intended result on powerpc? |
I intended for the issue to be just for the failing test cases. I investigated the issue and unrolling is working as expected. But at different unroll count and different interleave with an unchecked add for incrementing the index indvar. On PPC we get unroll factor of 8 with vector width of 2. Feel free to commit this, but leave the issue open for now and I can make the tests a bit more generic. |
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.
Thanks for doing this. But please leave the issue open, and I can update the testcase to be a bit more generic.
I think it's generally frowned upon to have test cases without From the linked issue, I think it is working on PPC, but it's just unrolling much more aggressively than here. At a glance it looks like it's unrolling 8 iterations at a time instead of just 2 like aarch64 and x86_64. I wonder if there's a |
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/157/builds/18291 Here is the relevant piece of the build log for the reference
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/89/builds/15256 Here is the relevant piece of the build log for the reference
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/140/builds/15622 Here is the relevant piece of the build log for the reference
|
The |
Interesting. Why are we not able to output as far as LLVM IR without having the backend registered? |
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/80/builds/9242 Here is the relevant piece of the build log for the reference
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/29/builds/9549 Here is the relevant piece of the build log for the reference
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/50/builds/9448 Here is the relevant piece of the build log for the reference
|
…n working targets" (#124536) Reverts llvm/llvm-project#123939
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/201/builds/2242 Here is the relevant piece of the build log for the reference
|
If -funroll-loops tests are not restricted to specific targets the tests may behave differently based on the host platform. This patch restricts the tests to aarch64 and x86_64, and removes the PowerPC XFAIL.