Improvements for cross-module-optimization

[eeckstein]

• Include small non-generic functions for serializaion
• serialize initializer of global variables: so that global let variables can be constant propagated across modules

rdar://problem/60696510

For details see the commit messages

[eeckstein]

@swift-ci test

[eeckstein]

@swift-ci benchmark

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
SuffixAnySequence	258	2052	+695.3%	0.13x
DropLastAnySequence	635	1924	+203.0%	0.33x
ObjectiveCBridgeStubFromNSDateRef	4720	5760	+22.0%	0.82x (?)
UTF8Decode_InitFromCustom_contiguous_ascii_as_ascii	390	441	+13.1%	0.88x (?)
CharacterLiteralsLarge	97	108	+11.3%	0.90x (?)
DropFirstSequence	53	59	+11.3%	0.90x
DropFirstSequenceLazy	53	59	+11.3%	0.90x
Improvement	OLD	NEW	DELTA	RATIO
EqualSubstringSubstring	41	38	-7.3%	1.08x
LessSubstringSubstring	41	38	-7.3%	1.08x
EqualSubstringSubstringGenericEquatable	41	38	-7.3%	1.08x
EqualSubstringString	41	38	-7.3%	1.08x (?)
LessSubstringSubstringGenericComparable	41	38	-7.3%	1.08x
EqualStringSubstring	42	39	-7.1%	1.08x (?)
NSStringConversion.MutableCopy.Rebridge	996	929	-6.7%	1.07x (?)

Code size: -O

Regression	OLD	NEW	DELTA	RATIO
Suffix.o	22189	22581	+1.8%	0.98x

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DropLastAnySequence	731	2314	+216.6%	0.32x
SuffixAnySequence	800	2169	+171.1%	0.37x
PrefixAnyCollection	182	217	+19.2%	0.84x
SuffixAnyCollection	65	76	+16.9%	0.86x
PrefixWhileAnyCollection	235	269	+14.5%	0.87x
DropFirstSequence	53	59	+11.3%	0.90x
DropFirstSequenceLazy	53	59	+11.3%	0.90x
CharacterLiteralsLarge	100	111	+11.0%	0.90x
ObjectiveCBridgeStubDateAccess	257	285	+10.9%	0.90x
PrefixAnySeqCRangeIterLazy	176	194	+10.2%	0.91x
PrefixAnySeqCntRangeLazy	176	194	+10.2%	0.91x
SuffixCountableRangeLazy	11	12	+9.1%	0.92x (?)
Array2D	7520	8112	+7.9%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
DropFirstAnySeqCRangeIter	61029	182	-99.7%	335.32x
DropFirstAnySeqCntRange	61024	182	-99.7%	335.29x
PrefixAnySeqCntRange	46728	175	-99.6%	267.02x
PrefixAnySeqCRangeIter	46321	193	-99.6%	240.00x
DropWhileSequenceLazy	140	105	-25.0%	1.33x
DropWhileAnySequenceLazy	2524	2125	-15.8%	1.19x (?)
ObjectiveCBridgeStubFromNSDate	7320	6320	-13.7%	1.16x (?)
PrefixAnySequenceLazy	1725	1581	-8.3%	1.09x (?)
SuffixArrayLazy	12	11	-8.3%	1.09x
SuffixArray	12	11	-8.3%	1.09x (?)
EqualSubstringSubstring	41	38	-7.3%	1.08x (?)
EqualStringSubstring	41	38	-7.3%	1.08x (?)
EqualSubstringSubstringGenericEquatable	41	38	-7.3%	1.08x (?)
EqualSubstringString	41	38	-7.3%	1.08x (?)
LessSubstringSubstring	42	39	-7.1%	1.08x (?)
LessSubstringSubstringGenericComparable	42	39	-7.1%	1.08x

Code size: -Osize

Improvement	OLD	NEW	DELTA	RATIO
DropFirst.o	18320	17226	-6.0%	1.06x
Prefix.o	18053	17723	-1.8%	1.02x

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
StrComplexWalk	7560	8250	+9.1%	0.92x
Improvement	OLD	NEW	DELTA	RATIO
EqualSubstringSubstring	49	45	-8.2%	1.09x (?)

Code size: -swiftlibs

Improvement	OLD	NEW	DELTA	RATIO
libswiftDarwin.dylib	32768	28672	-12.5%	1.14x

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 Pro
  Model Identifier: MacPro6,1
  Processor Name: 12-Core Intel Xeon E5
  Processor Speed: 2.7 GHz
  Number of Processors: 1
  Total Number of Cores: 12
  L2 Cache (per Core): 256 KB
  L3 Cache: 30 MB
  Memory: 64 GB

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - a7425c1

[eeckstein]

@swift-ci test linux

[compnerd]

Hey @eeckstein - this seems to have been checked in with a failing windows test. The CI actually caught that, and the bots are currently red. Could you please look into that or revert this please?

[eeckstein]

@compnerd Sorry. That's a reminder to always look at the windows build log, too. #32520 will fix it. Unfortunately PR testing is currently broken, so I cannot merge it :-(