[ownership] Change guaranteed args from transformation terminators to not require end_borrows. #29600

gottesmm · 2020-02-02T20:59:46Z

No description provided.

gottesmm · 2020-02-02T22:21:32Z

Lets see what source compat says.

gottesmm · 2020-02-02T22:21:39Z

@swift-ci test source compatibility

gottesmm · 2020-02-02T22:21:46Z

@swift-ci test linux platform

gottesmm · 2020-02-02T22:25:29Z

Hmm... error already came up when building stdlib in release locally. These are going to fail.

gottesmm · 2020-02-02T22:55:10Z

@swift-ci test source compatibility

gottesmm · 2020-02-02T22:55:14Z

@swift-ci test source compatibility

gottesmm · 2020-02-02T22:55:20Z

@swift-ci test source compatibility

gottesmm · 2020-02-02T22:55:25Z

@swift-ci test source compatibility

swift-ci · 2020-02-02T23:00:05Z

Build failed
Swift Test Linux Platform
Git Sha - 917dd4dafdaa373f53b5ed9c604eaab1b113582b

@owned

… not require end_borrows. For those who are unaware, a transformation terminator is a terminator like switch_enum/checked_cast_br that always dominate their successor blocks. Since they dominate their successor blocks by design and transform their input into the args form, we can validate that they obey guaranteed ownership semantics just like a forwarding instruction. Beyond removing unnecessary code bloat, this also makes it significantly more easier to optimize/work with transformation terminators when converting @owned -> @guaranteed since we do not need to find end_borrow points when the owned value is consumed. <rdar://problem/59097063>

gottesmm · 2020-02-03T01:05:10Z

@swift-ci test

gottesmm · 2020-02-03T01:05:13Z

@swift-ci test

gottesmm · 2020-02-03T01:05:19Z

@swift-ci benchmark

gottesmm · 2020-02-03T01:05:50Z

@swift-ci benchmark

gottesmm · 2020-02-03T01:05:54Z

@swift-ci test

gottesmm · 2020-02-03T01:06:02Z

@swift-ci test source compatibility

gottesmm · 2020-02-03T01:06:08Z

@swift-ci test source compatibility

gottesmm · 2020-02-03T01:06:17Z

@swift-ci test source compatibility

gottesmm · 2020-02-03T01:06:24Z

@swift-ci test windows platform

gottesmm · 2020-02-03T01:06:38Z

@swift-ci test windows platform

gottesmm · 2020-02-03T01:06:49Z

@swift-ci test windows platform

swift-ci · 2020-02-03T01:34:21Z

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	5310	5735	+8.0%	0.93x (?)

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	250	142	-43.2%	1.76x
PrefixAnySeqCntRangeLazy	18	14	-22.2%	1.29x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DataSetCountSmall	80	87	+8.7%	0.92x

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	222	100	-55.0%	2.22x
DataAccessBytesMedium	56	50	-10.7%	1.12x (?)
Data.init.Sequence.64kB.Count.RE.I	55	50	-9.1%	1.10x (?)
Data.append.Sequence.64kB.Count.RE.I	33	30	-9.1%	1.10x (?)
Data.init.Sequence.809B.Count.RE	103	95	-7.8%	1.08x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
DataAppendDataSmallToSmall	2880	3280	+13.9%	0.88x (?)

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	294	162	-44.9%	1.81x
DataCopyBytesSmall	130	115	-11.5%	1.13x

Code size: -swiftlibs

Improvement	OLD	NEW	DELTA	RATIO
libswiftStdlibUnittest.dylib	335872	327680	-2.4%	1.02x

How to read the data

The 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
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac mini
  Model Identifier: Macmini8,1
  Processor Name: Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

gottesmm · 2020-02-03T01:34:26Z

@swift-ci test windows platform

gottesmm · 2020-02-03T01:34:56Z

Hmmm... I have no clue why that benchmark improved so much.

gottesmm · 2020-02-03T01:35:15Z

@swift-ci benchmark

swift-ci · 2020-02-03T01:54:33Z

Performance: -O

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	250	142	-43.2%	1.76x
PolymorphicCalls	15	14	-6.7%	1.07x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DataSetCountSmall	80	87	+8.7%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	222	100	-55.0%	2.22x
Data.init.Sequence.64kB.Count.RE.I	56	50	-10.7%	1.12x (?)
Data.append.Sequence.64kB.Count.RE.I	33	30	-9.1%	1.10x (?)
DataAccessBytesMedium	54	50	-7.4%	1.08x (?)
Data.init.Sequence.64kB.Count.RE	54	50	-7.4%	1.08x (?)

Code size: -Osize

Performance: -Onone

Improvement	OLD	NEW	DELTA	RATIO
DataCopyBytesMedium	284	162	-43.0%	1.75x
DataCopyBytesSmall	130	115	-11.5%	1.13x (?)

Code size: -swiftlibs

Improvement	OLD	NEW	DELTA	RATIO
libswiftStdlibUnittest.dylib	335872	327680	-2.4%	1.02x

How to read the data

The 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
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac mini
  Model Identifier: Macmini8,1
  Processor Name: Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

gottesmm force-pushed the pr-d1b175b6bba114b57173be25ad701464ec91e4b9 branch from 22b4f99 to 917dd4d Compare February 2, 2020 22:20

gottesmm force-pushed the pr-d1b175b6bba114b57173be25ad701464ec91e4b9 branch from 917dd4d to 7b26a62 Compare February 2, 2020 22:30

gottesmm force-pushed the pr-d1b175b6bba114b57173be25ad701464ec91e4b9 branch from 7b26a62 to 6c5c853 Compare February 3, 2020 01:05

gottesmm changed the title ~~[DRAFT][ownership] Change guaranteed args from transformation terminators to not require end_borrows.~~ [ownership] Change guaranteed args from transformation terminators to not require end_borrows. Feb 3, 2020

gottesmm requested a review from atrick February 3, 2020 01:06

gottesmm mentioned this pull request Feb 3, 2020

[semantic-arc-opts] Convert @owned -> @guaranteed args of transforming terminators when fed by load [copy], copy_value. #29605

Merged

gottesmm merged commit 643227e into swiftlang:master Feb 3, 2020

gottesmm deleted the pr-d1b175b6bba114b57173be25ad701464ec91e4b9 branch February 3, 2020 07:39

[ownership] Change guaranteed args from transformation terminators to not require end_borrows. #29600

[ownership] Change guaranteed args from transformation terminators to not require end_borrows. #29600

Uh oh!

Conversation

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 2, 2020

Uh oh!

swift-ci commented Feb 2, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

swift-ci commented Feb 3, 2020

Performance: -O

Code size: -O

Performance: -Osize

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

gottesmm commented Feb 3, 2020

Uh oh!

swift-ci commented Feb 3, 2020

Performance: -O

Code size: -O

Performance: -Osize

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Uh oh!

Uh oh!