[CI][UR] Merge latest updates in UR benchmarking suite #17545
Closed
+1,885
−877
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
This patch improves numerous aspects on how the benchmarking results are visualized: - rewrites the way HTML charts are generated, using a library (Chart.js) that's both easier to use and more visually pleasing. The new HTML page also now decouples data from the HTML itself, leading to faster load times and the ability to fetch data from remote sources. - The markdown output now contains a failures section that lists all benchmarks that failed for a given run. This will be a helpful for developers during PR testing. - Benchmarks can now have description that's displayed on the page. - And many more minor improvements.
…to unify-benchmark-ci
This reverts commit f79bbbf.
... and add presets to more easily
…to unify-benchmark-ci
…to unify-benchmark-ci
On PRs based on main, the scripts location is "old" and not accesible. Pick location based on the dir existance. Step 'gather info' is in a 'weird' location, so solve it with 2 tries to execute the script.
uditagarwal97
pushed a commit
that referenced
this pull request
Mar 20, 2025
Scripts enabling CI for benchmarking has gotten pretty big and complicated: #17545 Most of this code were developed by the UR team, and thus it no longer makes sense to ask dpcpp-devops-reviewers for PR approvals. This PR updates CODEOWNERS to ask llvm-reviewers-benchmarking for benchmarking scripts instead. **Note:** The current path here puts `/devops/scripts/benchmarks` under llvm-reviewers-benchmarking: This path currently does not exist, but is chosen on purpose in anticipation for #17545. The folder `/devops/scripts/benchmarking` will be removed in the future as a part of the effort to unify SYCL and UR benchmarking CI infrastructure; See #17229
|
I'm realizing this is still quite a large PR, and people don't want to read or approve 1.8k lines of code 😅 Thus, I am going to further split this PR into smaller chunks: Phase 1 (PRs that needs to be merged in series):
Phase 2 (Tentatively; PRs that can be filed independently of eachother):
I'm not sure if this is too much splitting, but hopefully it preserves everybody's sanities |
martygrant
pushed a commit
that referenced
this pull request
Mar 24, 2025
Given that #17229 is massive, I am splitting the PR into smaller individual PRs such that reviewers would have an easier time. This PR prepares for changes in #17229 by moving the UR benchmarking scripts into devops. For more context: A previous attempt at splitting is here #17545, but even this is way too big, and thus I'm splitting it into chunks. This is a non-functional change, as these scripts not currently used anywhere right now... hopefully soon though!
RossBrunton
pushed a commit
to oneapi-src/unified-runtime
that referenced
this pull request
Mar 24, 2025
Given that intel/llvm#17229 is massive, I am splitting the PR into smaller individual PRs such that reviewers would have an easier time. This PR prepares for changes in intel/llvm#17229 by moving the UR benchmarking scripts into devops. For more context: A previous attempt at splitting is here intel/llvm#17545, but even this is way too big, and thus I'm splitting it into chunks. This is a non-functional change, as these scripts not currently used anywhere right now... hopefully soon though!
This was referenced Mar 25, 2025
martygrant
pushed a commit
that referenced
this pull request
Mar 26, 2025
…rks (#17617) This PR partially merges changes introduced in #17229. Specifically, it merges changes related to: - Introducing metadata in the result - Tweaks to the benchmarks itself: benchmarks ran, parameters used to run, what is enabled/disabled, etc - Enables/Disables benchmarks we are interested/no longer interested in - Disables non-working benchmarks - Bumps commit hashes used to fetch each benchmark - New command line arguments, such as `--build-jobs` and `--redownload` (to re-fetch the benchmarks) **Note:** I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) I did not try to clean up the commit history. Most of these changes are by Piotr anyway, I wanted to make sure he was properly credited for the changes. Updates for current splitting effort of #17229: #17545 (comment) --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
This was referenced Mar 26, 2025
martygrant
pushed a commit
that referenced
this pull request
Mar 27, 2025
In continuation of the effort outlined in #17545 (comment), this PR merges further changes introduced in #17229. Specifically, it merges @pbalcer's changes for adding the ability to run different benchmarking presets **Note:** I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
martygrant
pushed a commit
that referenced
this pull request
Mar 27, 2025
…ntation (#17662) In continuation of the effort outlined in #17545 (comment), this PR merges further changes introduced in #17229. Specifically, it merges changes related to markdown output generation intended for creating summaries on user PRs. Note: I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
martygrant
pushed a commit
that referenced
this pull request
Mar 28, 2025
In continuation of the effort outlined in #17545 (comment), this PR merges further changes introduced in #17229. Specifically, it merges @pbalcer's new HTML benchmarking results dashboard implementation, which now loads html data from remote sources in order to facilitate bringing up the benchmarking dashboard in intel.github.io in the future. **Note:** I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) I did not try to clean up the commit history. Most of these changes are by Piotr anyway, I wanted to make sure he was properly credited for the changes. --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
mateuszpn
pushed a commit
to mateuszpn/llvm
that referenced
this pull request
Mar 31, 2025
…#17659) In continuation of the effort outlined in intel#17545 (comment), this PR merges further changes introduced in intel#17229. Specifically, it merges @pbalcer's new HTML benchmarking results dashboard implementation, which now loads html data from remote sources in order to facilitate bringing up the benchmarking dashboard in intel.github.io in the future. **Note:** I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) I did not try to clean up the commit history. Most of these changes are by Piotr anyway, I wanted to make sure he was properly credited for the changes. --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
ggojska
pushed a commit
to ggojska/llvm
that referenced
this pull request
Apr 7, 2025
…#17659) In continuation of the effort outlined in intel#17545 (comment), this PR merges further changes introduced in intel#17229. Specifically, it merges @pbalcer's new HTML benchmarking results dashboard implementation, which now loads html data from remote sources in order to facilitate bringing up the benchmarking dashboard in intel.github.io in the future. **Note:** I am relying on this PR having its commits squashed during merge (which should be the default behavior for intel/llvm) I did not try to clean up the commit history. Most of these changes are by Piotr anyway, I wanted to make sure he was properly credited for the changes. --------- Co-authored-by: Piotr Balcer <[email protected]> Co-authored-by: Łukasz Stolarczuk <[email protected]>
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.
This PR:
/devops, where the SYCL CI scripts resideThis PR serves as a stepping stone for combining the SYCL benchmarking CI with the UR benchmarking CI: Given that #17229 is massive, I am splitting that PR into smaller individual PRs so that reviewers have an easier time.
Note: I am depending on this commit to be merged with commit squashing turns on, otherwise the commit history is rather unsavory.
Note 2: Much of the benchmarking changes were already reviewed here: #17475