RFR: 8365205: C2: Optimize popcount value computation using knownbits [v11]

[Chen Liang]

On Fri, 19 Sep 2025 20:44:54 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>> This patch optimizes PopCount value transforms using KnownBits information.
>> Following are the results of the micro-benchmark included with the patch
>> 
>> 
>> 
>> System: 13th Gen Intel(R) Core(TM) i3-1315U
>> 
>> Baseline:
>> Benchmark                                      Mode  Cnt       Score   Error  Units
>> PopCountValueTransform.LogicFoldingKerenLong  thrpt    2  215460.670          ops/s
>> PopCountValueTransform.LogicFoldingKerenlInt  thrpt    2  294014.826          ops/s
>> 
>> Withopt:
>> Benchmark                                      Mode  Cnt       Score   Error  Units
>> PopCountValueTransform.LogicFoldingKerenLong  thrpt    2  389978.082          ops/s
>> PopCountValueTransform.LogicFoldingKerenlInt  thrpt    2  417261.583          ops/s
>> 
>> 
>> Kindly review and share your feedback.
>> 
>> Best Regards,
>> Jatin
>
> Jatin Bhateja has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Update countbitsnode.cpp

A suggestion, we might do this as a later RFE.

src/hotspot/share/opto/countbitsnode.cpp line 124:

> 122: // From the definition of KnownBits, we know:
> 123: //   zeros: Indicates which bits must be 0: zeros[i]=1 -> t[i]=0
> 124: //   ones:  Indicates which bits must be 1: ones[i]=1 -> t[i]=1

I don't think we should duplicate the information available in rangeinference.hpp - we should enhance the documentation there, such as noting `~zeros` is the number where the maximum number of bits are set to 1, to help comprehension.

If we know `ones` is `min_ones` and `~zeros` is `max_ones`, we can easily derive that `pop_count(ones) <= pop_count(t) <= pop_count(~zeros)`

-------------

PR Review: https://git.openjdk.org/jdk/pull/27075#pullrequestreview-3269507322
PR Review Comment: https://git.openjdk.org/jdk/pull/27075#discussion_r2380355747