[DNM] disable assign/partial store as a test. #21918
Closed
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
|
@swift-ci smoke benchmark |
Build comment file:Performance: -O
Code size: -O
Performance: -Osize
Code size: -Osize
Performance: -Onone
Code size: -swiftlibs
How to read the dataThe tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.If you see any unexpected regressions, you should consider fixing the Noise: Sometimes the performance results (not code size!) contain false Hardware Overview |
gottesmm
force-pushed
the
pr-75a803f814bbe0a7bd5b5533baf879c9e1c806b3
branch
from
4c858de to
85ee34f
Compare
|
@swift-ci smoke benchmark |
|
@swift-ci test |
|
Build failed |
|
Build failed |
|
Build failed |
Build comment file:Performance: -Onone
How to read the dataThe tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.If you see any unexpected regressions, you should consider fixing the Noise: Sometimes the performance results (not code size!) contain false Hardware Overview |
… stops allocations from being removed. PartialStore and Assign in PMO today both are allowed and do not prevent an allocation from being removed. This is incorrect with ownership since in both cases if we have a non-trivial type, we would need to make sure that the value being overwritten is destroyed. Clearly, with ownership given that we do not know how to handle assigns in the aforementioned way, we can not remove allocation with ownership. When I make the optimization more conservative by changing assigns to stop allocations from being removed, I see no difference in the tests/benchmarks/etc. So it makes sense to make the change now. The reason to remove PartialStore is that: 1. It is actually a vestigal remnant of Definite Init code in Predictable Mem Opts. If you look in Definite Init, understanding PartialStores is a very important part of the algorithm for diagnosing partially initialized memory locations. In Predictable Mem Opts it only controlled whether or not we could eliminate an allocation. 2. PartialStore assumes that values are always initialized completely so a partial store must be an assign. Despite that the two places it was used in the code was a place where we assumed that SILGen was always going to always have an init anyways (meaning we produced a partial store without need) or in a case where if we do not have an init, we can just treat it like an assign (the copy_addr) case.
gottesmm
force-pushed
the
pr-75a803f814bbe0a7bd5b5533baf879c9e1c806b3
branch
from
85ee34f to
99d86f5
Compare
gottesmm
changed the title
gottesmm
added a commit
to gottesmm/swift
that referenced
this pull request
Jan 20, 2019
…ign of the dest rather than like a non-trivial type. PMO uses InitOrAssign for trivially typed things and Init/Assign for non-trivial things, so I think this was an oversight from a long time ago. There is actually no /real/ effect on the code today since after exploding the copy_addr, the store will still be used to produce the right available value and since for stores, init/assign/initorassign all result in allocations being removed. Once though I change assign to not allow for allocation removal (the proper way to model this), without this change, certain trivial allocations will no longer be removed, harming perf as seen via the benchmarking run on the bots in swiftlang#21918.
|
Closing this. It served its purpose. |
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.
Just a test.