Re: RFR: 8196106: Support nested infinite or recursive flat mapped streams [v2]
On Tue, 9 Apr 2024 10:07:46 GMT, Viktor Klang wrote: >> This PR implements Gatherer-inspired encoding of `flatMap` that shows that >> it is both competitive performance-wise as well as improve correctness. >> >> Below is the performance of `Stream::flatMap` (for reference types): >> >> Before this PR: >> >> Benchmark(size) Mode CntScore Error Units >> FlatMap.par_array10 thrpt 12 294008,937 ? 54369,110 ops/s >> FlatMap.par_array 100 thrpt 1262411,229 ? 14868,119 ops/s >> FlatMap.par_array 1000 thrpt 12 8263,821 ? 452,622 ops/s >> FlatMap.par_iterate 10 thrpt 1223029,978 ? 4274,449 ops/s >> FlatMap.par_iterate 100 thrpt 1210532,907 ? 321,694 ops/s >> FlatMap.par_iterate1000 thrpt 12 981,571 ? 135,270 ops/s >> FlatMap.seq_array10 thrpt 12 2955648,495 ? 32539,142 ops/s >> FlatMap.seq_array 100 thrpt 1241851,009 ? 377,546 ops/s >> FlatMap.seq_array 1000 thrpt 12 1740,281 ? 1229,974 ops/s >> FlatMap.seq_iterate 10 thrpt 12 321727,690 ? 5149,356 ops/s >> FlatMap.seq_iterate 100 thrpt 12 8437,198 ?56,635 ops/s >> FlatMap.seq_iterate1000 thrpt 12 76,994 ? 0,965 ops/s >> >> >> After this PR: >> >> >> Benchmark(size) Mode CntScoreError Units >> FlatMap.par_array10 thrpt 12 283350,051 ? 35567,223 ops/s >> FlatMap.par_array 100 thrpt 1253846,906 ? 19241,913 ops/s >> FlatMap.par_array 1000 thrpt 12 8230,909 ?156,362 ops/s >> FlatMap.par_iterate 10 thrpt 1226328,500 ? 5411,401 ops/s >> FlatMap.par_iterate 100 thrpt 1210470,862 ?249,991 ops/s >> FlatMap.par_iterate1000 thrpt 12 986,511 ?224,050 ops/s >> FlatMap.seq_array10 thrpt 12 5654826,565 ? 27317,453 ops/s >> FlatMap.seq_array 100 thrpt 12 187929,786 ?542,787 ops/s >> FlatMap.seq_array 1000 thrpt 12 2385,346 ? 9,827 ops/s >> FlatMap.seq_iterate 10 thrpt 12 812722,403 ? 160500,399 ops/s >> FlatMap.seq_iterate 100 thrpt 1213542,472 ?118,769 ops/s >> FlatMap.seq_iterate1000 thrpt 12 157,056 ? 1,814 ops/s >> >> >> For streams of size 100k, the following numbers are interesting: >> >> Before this PR: >> >> Benchmark(size) Mode Cnt ScoreError Units >> FlatMap.par_array10 thrpt 12 0,325 ? 0,004 ops/s >> FlatMap.par_iterate 10 thrpt 12 0,106 ? 0,008 o... > > Viktor Klang has updated the pull request incrementally with one additional > commit since the last revision: > > Updating copyright year src/java.base/share/classes/java/util/stream/DoublePipeline.java line 280: > 278: result.sequential().allMatch(this); > 279: else > 280: > result.sequential().forEach(sink::accept); I think that might create a new double consumer instance for every input element. Alternatively you can compute and cache it as a field, replacing `shorts` and use a `null` check. - PR Review Comment: https://git.openjdk.org/jdk/pull/18625#discussion_r1557995257
Re: RFR: 8196106: Support nested infinite or recursive flat mapped streams [v2]
On Tue, 9 Apr 2024 10:04:44 GMT, Viktor Klang wrote:
>> This PR implements Gatherer-inspired encoding of `flatMap` that shows that
>> it is both competitive performance-wise as well as improve correctness.
>>
>> Below is the performance of `Stream::flatMap` (for reference types):
>>
>> Before this PR:
>>
>>
>> Benchmark(size) Mode CntScore Error Units
>> FlatMap.par_array10 thrpt 12 294008,937 ? 54369,110 ops/s
>> FlatMap.par_array 100 thrpt 1262411,229 ? 14868,119 ops/s
>> FlatMap.par_array 1000 thrpt 12 8263,821 ? 452,622 ops/s
>> FlatMap.par_iterate 10 thrpt 1223029,978 ? 4274,449 ops/s
>> FlatMap.par_iterate 100 thrpt 1210532,907 ? 321,694 ops/s
>> FlatMap.par_iterate1000 thrpt 12 981,571 ? 135,270 ops/s
>> FlatMap.seq_array10 thrpt 12 2955648,495 ? 32539,142 ops/s
>> FlatMap.seq_array 100 thrpt 1241851,009 ? 377,546 ops/s
>> FlatMap.seq_array 1000 thrpt 12 1740,281 ? 1229,974 ops/s
>> FlatMap.seq_iterate 10 thrpt 12 321727,690 ? 5149,356 ops/s
>> FlatMap.seq_iterate 100 thrpt 12 8437,198 ?56,635 ops/s
>> FlatMap.seq_iterate1000 thrpt 12 76,994 ? 0,965 ops/s
>>
>>
>> After this PR:
>>
>>
>> Benchmark(size) Mode CntScoreError Units
>> FlatMap.par_array10 thrpt 12 283350,051 ? 35567,223 ops/s
>> FlatMap.par_array 100 thrpt 1253846,906 ? 19241,913 ops/s
>> FlatMap.par_array 1000 thrpt 12 8230,909 ?156,362 ops/s
>> FlatMap.par_iterate 10 thrpt 1226328,500 ? 5411,401 ops/s
>> FlatMap.par_iterate 100 thrpt 1210470,862 ?249,991 ops/s
>> FlatMap.par_iterate1000 thrpt 12 986,511 ?224,050 ops/s
>> FlatMap.seq_array10 thrpt 12 5654826,565 ? 27317,453 ops/s
>> FlatMap.seq_array 100 thrpt 12 187929,786 ?542,787 ops/s
>> FlatMap.seq_array 1000 thrpt 12 2385,346 ? 9,827 ops/s
>> FlatMap.seq_iterate 10 thrpt 12 812722,403 ? 160500,399 ops/s
>> FlatMap.seq_iterate 100 thrpt 1213542,472 ?118,769 ops/s
>> FlatMap.seq_iterate1000 thrpt 12 157,056 ? 1,814 ops/s
>
> Viktor Klang has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Updating copyright year
src/java.base/share/classes/java/util/stream/DoublePipeline.java line 268:
> 266: class FlatMap implements Sink.OfDouble, DoublePredicate {
> 267: boolean cancel;
> 268: private final boolean shorts =
> isShortCircuitingPipeline();
If there's a known, short-circuiting, operation in the pipeline, we have to use
`allMatch` but in the case where we know that there's nothing which could
short-circuit we can go down a fast-path to `forEach`.
-
PR Review Comment: https://git.openjdk.org/jdk/pull/18625#discussion_r1557369039
Re: RFR: 8196106: Support nested infinite or recursive flat mapped streams [v2]
> This PR implements Gatherer-inspired encoding of `flatMap` that shows that it > is both competitive performance-wise as well as improve correctness. > > Below is the performance of `Stream::flatMap` (for reference types): > > Before this PR: > > > Benchmark(size) Mode CntScore Error Units > FlatMap.par_array10 thrpt 12 294008,937 ? 54369,110 ops/s > FlatMap.par_array 100 thrpt 1262411,229 ? 14868,119 ops/s > FlatMap.par_array 1000 thrpt 12 8263,821 ? 452,622 ops/s > FlatMap.par_iterate 10 thrpt 1223029,978 ? 4274,449 ops/s > FlatMap.par_iterate 100 thrpt 1210532,907 ? 321,694 ops/s > FlatMap.par_iterate1000 thrpt 12 981,571 ? 135,270 ops/s > FlatMap.seq_array10 thrpt 12 2955648,495 ? 32539,142 ops/s > FlatMap.seq_array 100 thrpt 1241851,009 ? 377,546 ops/s > FlatMap.seq_array 1000 thrpt 12 1740,281 ? 1229,974 ops/s > FlatMap.seq_iterate 10 thrpt 12 321727,690 ? 5149,356 ops/s > FlatMap.seq_iterate 100 thrpt 12 8437,198 ?56,635 ops/s > FlatMap.seq_iterate1000 thrpt 12 76,994 ? 0,965 ops/s > > > After this PR: > > > Benchmark(size) Mode CntScoreError Units > FlatMap.par_array10 thrpt 12 283350,051 ? 35567,223 ops/s > FlatMap.par_array 100 thrpt 1253846,906 ? 19241,913 ops/s > FlatMap.par_array 1000 thrpt 12 8230,909 ?156,362 ops/s > FlatMap.par_iterate 10 thrpt 1226328,500 ? 5411,401 ops/s > FlatMap.par_iterate 100 thrpt 1210470,862 ?249,991 ops/s > FlatMap.par_iterate1000 thrpt 12 986,511 ?224,050 ops/s > FlatMap.seq_array10 thrpt 12 5654826,565 ? 27317,453 ops/s > FlatMap.seq_array 100 thrpt 12 187929,786 ?542,787 ops/s > FlatMap.seq_array 1000 thrpt 12 2385,346 ? 9,827 ops/s > FlatMap.seq_iterate 10 thrpt 12 812722,403 ? 160500,399 ops/s > FlatMap.seq_iterate 100 thrpt 1213542,472 ?118,769 ops/s > FlatMap.seq_iterate1000 thrpt 12 157,056 ? 1,814 ops/s Viktor Klang has updated the pull request incrementally with one additional commit since the last revision: Updating copyright year - Changes: - all: https://git.openjdk.org/jdk/pull/18625/files - new: https://git.openjdk.org/jdk/pull/18625/files/e31c764f..3ff40739 Webrevs: - full: https://webrevs.openjdk.org/?repo=jdk&pr=18625&range=01 - incr: https://webrevs.openjdk.org/?repo=jdk&pr=18625&range=00-01 Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod Patch: https://git.openjdk.org/jdk/pull/18625.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/18625/head:pull/18625 PR: https://git.openjdk.org/jdk/pull/18625
