[benchmark] Janitor Duty: Final Sweep

[palimondo]

Hopefully last of the cleanups…

[palimondo]

@swift-ci please benchmark

[swift-ci]

Performance: -O

TEST	OLD	NEW	DELTA	RATIO
Improvement
RemoveWhereQuadraticInts	480	98	-79.6%	4.90x
RemoveWhereQuadraticStrings	1134	316	-72.1%	3.59x

Code size: -O

TEST	OLD	NEW	DELTA	RATIO
Regression
RemoveWhere.o	24391	24983	+2.4%	0.98x

Performance: -Osize

TEST	OLD	NEW	DELTA	RATIO
Improvement
RemoveWhereQuadraticInts	484	96	-80.2%	5.04x
RemoveWhereQuadraticStrings	1133	319	-71.8%	3.55x

Code size: -Osize

TEST	OLD	NEW	DELTA	RATIO
Regression
RemoveWhere.o	24118	24486	+1.5%	0.98x

Performance: -Onone

TEST	OLD	NEW	DELTA	RATIO
Improvement
RemoveWhereQuadraticStrings	9848	4634	-52.9%	2.13x
RemoveWhereQuadraticInts	8861	4284	-51.7%	2.07x

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

[palimondo]

Hmm… I was trying to lower the workload of RemoveWhereQuadratic because on my machine the runtimes on Ints and Strings variants were simply egregious:

TEST	MIN(μs)
RemoveWhereQuadraticInts	10397
RemoveWhereQuadraticStrings	25880

...but from the test benchmark above I see that these run much better on modern hardware:

TEST	OLD	NEW	DELTA	RATIO
RemoveWhereQuadraticInts	480	98	-79.6%	4.90x
RemoveWhereQuadraticStrings	1134	316	-72.1%	3.59x

I'm unable to use legacyFactor here, because there is non-linear relationship between the workload size and resulting runtime. I think the performance of this benchmark is heavily dependent on L1 and L2 cache size. I'm going to leave it as is.

[CodaFi]

@swift-ci test

[CodaFi]

@swift-ci benchmark

[swift-ci]

Performance: -O

Improvement	OLD	NEW	DELTA	RATIO
DataAppendDataSmallToSmall	4200	3840	-8.6%	1.09x (?)

Code size: -O

Regression	OLD	NEW	DELTA	RATIO
InsertCharacter.o	5115	5195	+1.6%	0.98x

Performance: -Osize

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
CharacterPropertiesFetch	5150	5860	+13.8%	0.88x (?)
String.data.LargeUnicode	123	135	+9.8%	0.91x (?)

Code size: -swiftlibs

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: 8-Core Intel Xeon E5
  Processor Speed: 3 GHz
  Number of Processors: 1
  Total Number of Cores: 8
  L2 Cache (per Core): 256 KB
  L3 Cache: 25 MB
  Memory: 64 GB

[palimondo]

@eeckstein I seem to have forgotten about this one. Thank's for remininding me, @CodaFi. Please review.

[eeckstein]

lgtm