Repository navigation
Report actual instead of requested iterations in memory results - #2322
Merged
Merged
Conversation
Contributor
Author
|
I could make it a gtest instead, and use EDIT: converted to a gtest. Bazel needs no updates because of the file matching logic. |
nicholasjng
force-pushed
the
push-oqskqzymwzkw
branch
from
October 7, 2026 15:43
3916023 to
60f538a
Compare
LebedevRI
reviewed
Oct 27, 2026
LebedevRI
requested changes
Oct 27, 2026
LebedevRI
reviewed
Oct 27, 2026
LebedevRI
reviewed
Oct 28, 2026
Collaborator
There was a problem hiding this comment.
Please drop this argument then.
LebedevRI
reviewed
Oct 28, 2026
LebedevRI
left a comment
LebedevRI
left a comment
Collaborator
There was a problem hiding this comment.
Ah, i see.
Can some of the explicit memory_iterations arguments be dropped then?
Previously, the run report contained the requested number of memory iterations. When using `KeepRunningBatch` with a batch size not dividing that number, the memory profiling overshoots, and additional iterations go unreported. This change sets the number of iterations that were actually run in the memory manager as `memory_result.memory_iterations`, and adds a regression test exercising the overshooting path via `KeepRunningBatch()`.
The guard against division by zero stays in place in the `allocs_per_iter` calculation, since memory_result.memory_iterations is initialized to zero, and can be left zero in a skipped memory manager pass.
nicholasjng
force-pushed
the
push-oqskqzymwzkw
branch
from
October 29, 2026 10:02
60f538a to
b57dd1d
Compare
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Previously, the run report contained the requested number of memory iterations. When using
KeepRunningBatchwith a batch size not dividing that number, the memory profiling overshoots, and additional iterations go unreported.This change sets the number of iterations that were actually run in the memory manager as
memory_result.memory_iterations, and adds a regression test exercising the overshooting path viaKeepRunningBatch().Found in experiments with
KeepRunningBatchand memory profiling in mew.disclaimer: I used AI for the creation of the unit test, reviewed it with the existing
profiler_manager_iterations_test.ccas reference, liked what I saw.