Reduce generics in Codable #31278

tbkka · 2020-04-24T19:23:21Z

In particular, types generic on the CodingKey produce a lot of runtime metadata.
Reducing the number of such types should help with some of the reported memory
bloat from Codable.

Resolves rdar://62620208

In particular, types generic on the CodingKey produce a lot of runtime metadata. Reducing the number of such types should help with some of the reported memory bloat from Codable.

tbkka · 2020-04-24T19:24:57Z

@swift-ci Please benchmark

stdlib/public/core/Codable.swift

swift-ci · 2020-04-24T19:55:46Z

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
UTF8Decode_InitDecoding	142	167	+17.6%	0.85x
UTF8Decode_InitFromCustom_contiguous	139	163	+17.3%	0.85x
ObjectiveCBridgeStringHash	72	81	+12.5%	0.89x
Set.subtracting.Empty.Box	8	9	+12.5%	0.89x (?)
OpenClose	56	61	+8.9%	0.92x (?)
UTF8Decode_InitFromCustom_noncontiguous	274	298	+8.8%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
MapReduce	219	197	-10.0%	1.11x (?)
RemoveWhereSwapInts	42	38	-9.5%	1.11x (?)
DistinctClassFieldAccesses	201	183	-9.0%	1.10x (?)
ArrayPlusEqualFiveElementCollection	5254	4847	-7.7%	1.08x (?)
ObjectiveCBridgeStubNSDateRefAccess	525	486	-7.4%	1.08x (?)
PrefixWhileSequence	192	178	-7.3%	1.08x (?)
ArraySetElement	305	284	-6.9%	1.07x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
UTF8Decode_InitDecoding	140	163	+16.4%	0.86x (?)
UTF8Decode_InitFromCustom_contiguous	143	164	+14.7%	0.87x
ObjectiveCBridgeStringHash	72	81	+12.5%	0.89x
StringComparison_ascii	348	384	+10.3%	0.91x (?)
Chars2	3200	3500	+9.4%	0.91x (?)
UTF8Decode_InitFromCustom_noncontiguous	276	301	+9.1%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
MapReduce	246	218	-11.4%	1.13x (?)
Set.isStrictSubset.Int.Empty	52	47	-9.6%	1.11x (?)
Set.isStrictSubset.Seq.Int.Empty	118	107	-9.3%	1.10x (?)
MapReduceAnyCollection	237	216	-8.9%	1.10x (?)
FlattenListLoop	2807	2584	-7.9%	1.09x (?)
ArraySetElement	283	262	-7.4%	1.08x (?)
Array2D	4544	4208	-7.4%	1.08x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
UTF8Decode_InitFromCustom_contiguous	151	176	+16.6%	0.86x
UTF8Decode_InitDecoding	148	172	+16.2%	0.86x (?)
ObjectiveCBridgeStringHash	72	81	+12.5%	0.89x
ArrayOfPOD	644	723	+12.3%	0.89x (?)
StringWalk	2760	3040	+10.1%	0.91x (?)
UTF8Decode_InitFromData	160	175	+9.4%	0.91x (?)
UTF8Decode_InitDecoding_ascii_as_ascii	216	236	+9.3%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
CharacterLiteralsLarge	374	346	-7.5%	1.08x (?)
Breadcrumbs.MutatedUTF16ToIdx.ASCII	15	14	-6.7%	1.07x (?)

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 mini
  Model Identifier: Macmini8,1
  Processor Name: 6-Core 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

tbkka · 2020-04-24T20:06:14Z

This didn't catch any perf regressions on the JSON encode/decode tests. I think we can dismiss the UTF8Decode checks above as unrelated noise, since they're not testing Codable/Decodable.

tbkka · 2020-04-24T20:06:24Z

@swift-ci Please benchmark

tbkka · 2020-04-24T20:29:55Z

stdlib/public/core/Codable.swift

@@ -3845,237 +3843,237 @@ internal class _KeyedDecodingContainerBase<Key: CodingKey> {
    fatalError("_KeyedDecodingContainerBase cannot be used directly.")
  }

-  internal var allKeys: [Key] {
+  internal var allKeys: [CodingKey] {


@jckarter I don't suppose you have any ideas for improving this one?

Hm. Do we have a way using standard library SPI that we could just store the reference to the array's buffer object here, and construct the Array<Key> shell around it in the getter?

If the SPI doesn't exist, maybe you could have this variable be of type _ArrayBridgeStorage and bitcast it back and forth to Array<K> in the methods that access it.

swift-ci · 2020-04-24T20:40:49Z

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
AnyHashableWithAClass	119000	129000	+8.4%	0.92x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubToNSDate2	550	640	+16.4%	0.86x (?)
AnyHashableWithAClass	119000	129000	+8.4%	0.92x (?)
PlistPerfEncode	494	535	+8.3%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
DictionaryBridgeToObjC_Access	961	828	-13.8%	1.16x (?)
FlattenListFlatMap	7657	6977	-8.9%	1.10x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
PlistPerfEncode	513	560	+9.2%	0.92x (?)
AnyHashableWithAClass	120500	130500	+8.3%	0.92x (?)
ObjectiveCBridgeFromNSArrayAnyObjectForced	6940	7500	+8.1%	0.93x (?)
ArrayOfPOD	1030	1110	+7.8%	0.93x (?)

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: 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

@jckarter

… generic and manually asserting that the types match. Based on a suggestion of @jckarter

tbkka · 2020-04-24T21:20:44Z

I implemented @jckarter's suggestion -- let's see if that addresses the perf regression here.

tbkka · 2020-04-24T21:20:51Z

@swift-ci Please benchmark

tbkka · 2020-04-24T21:55:48Z

@swift-ci Please benchmark

swift-ci · 2020-04-24T23:03:18Z

Performance: -O

Improvement	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	8468	6138	-27.5%	1.38x (?)
FlattenListLoop	4724	4012	-15.1%	1.18x (?)
PrefixAnyCollectionLazy	91267	84599	-7.3%	1.08x (?)
ObjectiveCBridgeFromNSSetAnyObjectToString	69000	64000	-7.2%	1.08x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	5429	6641	+22.3%	0.82x (?)
ObjectiveCBridgeStubToNSDate2	500	550	+10.0%	0.91x (?)

Improvement	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeFromNSArrayAnyObjectForced	4220	3780	-10.4%	1.12x (?)

Code size: -Osize

Performance: -Onone

Improvement	OLD	NEW	DELTA	RATIO
ArrayOfPOD	982	907	-7.6%	1.08x (?)

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

tbkka · 2020-04-27T16:47:36Z

@swift-ci Please test

swift-ci · 2020-04-27T17:30:39Z

Build failed
Swift Test OS X Platform
Git Sha - cc1bf62

tbkka · 2020-04-27T19:23:54Z

@swift-ci Please test macOS

tbkka · 2020-05-05T17:01:24Z

@swift-ci Please nominate

Explanation: Reduce memory requirements for Codable/Decodable support by making internal types non-generic
Scope: Changes to private classes within Codable/Decodable implementation
Radar: rdar://62620208
Risk: Possible performance impacts to Codable/Decodable users
Testing: Full CI, local testing, CI benchmarking

@jckarter

In particular, types generic on the CodingKey produce a lot of runtime metadata. Reducing the number of such types should help with some of the reported memory bloat from Codable. Based on a suggestion of @jckarter Resolves rdar://62620208

@jckarter

In particular, types generic on the CodingKey produce a lot of runtime metadata. Reducing the number of such types should help with some of the reported memory bloat from Codable. Based on a suggestion of @jckarter Resolves rdar://62620208

Reduce generics in Codable

de85d14

In particular, types generic on the CodingKey produce a lot of runtime metadata. Reducing the number of such types should help with some of the reported memory bloat from Codable.

jckarter reviewed Apr 24, 2020

View reviewed changes

stdlib/public/core/Codable.swift Outdated Show resolved Hide resolved

tbkka commented Apr 24, 2020

View reviewed changes

Avoid a pointless cast operation on each call by making the functions…

9beb09d

… generic and manually asserting that the types match. Based on a suggestion of @jckarter

Simplify naming a bit

cc1bf62

tbkka marked this pull request as ready for review April 27, 2020 22:34

tbkka merged commit 70b3f01 into swiftlang:master Apr 29, 2020

tbkka mentioned this pull request May 13, 2020

[5.3] Reduce Generic use in Codable #31772

Merged

tbkka deleted the tbkka-decodable branch October 16, 2020 00:33

benrimmington mentioned this pull request Sep 21, 2021

Added protocol CodingKeyRepresentable #34458

Merged

benrimmington mentioned this pull request Feb 27, 2023

[SE-NNNN] Simplify internals of Codable-related APIs #63172

Closed

Reduce generics in Codable #31278

Reduce generics in Codable #31278

Uh oh!

Conversation

tbkka commented Apr 24, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

Uh oh!

swift-ci commented Apr 24, 2020

Performance: -O

Code size: -O

Performance: -Osize

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

tbkka Apr 24, 2020

Choose a reason for hiding this comment

Uh oh!

jckarter Apr 24, 2020

Choose a reason for hiding this comment

Uh oh!

jckarter Apr 27, 2020

Choose a reason for hiding this comment

Uh oh!

swift-ci commented Apr 24, 2020

Performance: -O

Code size: -O

Performance: -Osize

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

tbkka commented Apr 24, 2020

Uh oh!

swift-ci commented Apr 24, 2020

Performance: -O

Code size: -O

Performance: -Osize

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Uh oh!

tbkka commented Apr 27, 2020

Uh oh!

swift-ci commented Apr 27, 2020

Uh oh!

tbkka commented Apr 27, 2020

Uh oh!

tbkka commented May 5, 2020

Uh oh!

Uh oh!

tbkka commented Apr 24, 2020 •

edited

Loading