Fix: [ATen] Add some missing moves #88514
Conversation
Skylion007
requested review from
digantdesai,
jerryzh168,
jianyuh,
kimishpatel,
salilsdesai and
z-a-f
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/88514
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 FailuresAs of commit 4631b3f: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
github-actions
bot
added
module: cpu
Skylion007
force-pushed
the
skylion007-aten-fix-move
branch
from
009d51f to
b6ec53d
Compare
Skylion007
force-pushed
the
skylion007-aten-fix-move
branch
from
b6ec53d to
136ee1e
Compare
Skylion007
force-pushed
the
skylion007-aten-fix-move
branch
from
496321a to
136ee1e
Compare
…ion007-aten-fix-move
…ion007-aten-fix-move
…007/pytorch into skylion007-aten-fix-move
Skylion007
force-pushed
the
skylion007-aten-fix-move
branch
from
d3b0d7b to
f139b52
Compare
|
@jgong5 Merged master to fix the flake, this PR is ready to be merged assuming all tests pass. |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 2 additional jobs have failed, first few of them are: trunk ,trunk / macos-12-py3-arm64 / test (default, 1, 2, macos-m1-12) Details for Dev Infra teamRaised by workflow job |
|
@ezyang Failure here seems to be another infra failure unrelated to this PR. |
|
Tracking here: #88968 |
|
@pytorchbot merge -f "infra failure only" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Related to pytorch#88512 , but for ATen. This should reduce a number of copies and inefficient atomic smart pointer increments. Pull Request resolved: pytorch#88514 Approved by: https://github.com/jgong5, https://github.com/ezyang
Applies some more missing std::move found by static analysis. This should improve performance and reduce unnecessary copies. This PR only targets ATen for now. And before you ask about the edits, std::move is optimal in a ternary operator as copy ellision cannot happen one. The best thing is probably rewriting it as an if else, but ultimately this should be performant enough. Followup to #88512 and #88514 Pull Request resolved: #89000 Approved by: https://github.com/ezyang
5 participants
Related to #88512 , but for ATen. This should reduce a number of copies and inefficient atomic smart pointer increments.
cc @jerryzh168 @jianyuh @raghuramank100 @jamesr66a @vkuzo @jgong5 @Xia-Weiwen @leslie-fang-intel @VitalyFedyunin @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10