Support collecting perf profiles remotely #275
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
If we're running tests on a remote machine with -DTEST_SUITE_REMOTE_HOST and generate a perf profile with `lit --param profile=perf`, currently the profile step will be run on the local machine and fail. This patch fixes this by running the profile step on the remote, as well as copying the generated profiles back to the host so tools like LNT can inspect them. There's two parts to this, first we need to run the remote module after the perf (and hpmcount) module so that we can wrap the profile step with ssh etc. Secondly, we need to scp over the generated .perf_data files like we also do for the profilegen.py module. However today this is also done as part of the profile step which we now need to start running over ssh. So to fix this, this adds a new "profile collect" step that runs after the profile step, but always runs on the local machine. I've tested this out on a remote build configured with `-DTEST_SUITE_REMOTE_HOST=... -DTEST_SUITE_USE_PERF=ON -DTEST_SUITE_PROFILE_GENERATE=ON -DTEST_SUITE_USE_IR_PGO=ON` and running the tests with `lit --param profile=perf`, and it seems to work.
lukel97
requested review from
asb,
MatzeB,
DavidSpickett,
omjavaid and
TamarChristinaArm
Fixes a "AttributeError: 'TestContext' object has no attribute 'profilefile'" when running tests without --param profile=perf
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Outdated suggestions cannot be applied.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.
If we're running tests on a remote machine with -DTEST_SUITE_REMOTE_HOST and generate a perf profile with
lit --param profile=perf, currently the profile step will be run on the local machine and fail.This patch fixes this by running the profile step on the remote, as well as copying the generated profiles back to the host so tools like LNT can inspect them.
There's two parts to this, first we need to run the remote module after the perf (and hpmcount) module so that we can wrap the profile step with ssh etc.
Secondly, we need to scp over the generated .perf_data files like we also do for the profilegen.py module. However today this is also done as part of the profile step which we now need to start running over ssh. So to fix this, this adds a new "profile collect" step that runs after the profile step, but always runs on the local machine.
I've tested this out on a remote build configured with
-DTEST_SUITE_REMOTE_HOST=... -DTEST_SUITE_USE_PERF=ON -DTEST_SUITE_PROFILE_GENERATE=ON -DTEST_SUITE_USE_IR_PGO=ONand running the tests withlit --param profile=perf, and it seems to work.