RFR: 8286279: [vectorapi] Only check index of masked lanes if offset is out of array boundary for masked store [v2]
Xiaohong Gong
xgong at openjdk.java.net
Fri May 13 07:06:52 UTC 2022
On Fri, 13 May 2022 01:35:40 GMT, Xiaohong Gong <xgong at openjdk.org> wrote:
>> Checking whether the indexes of masked lanes are inside of the valid memory boundary is necessary for masked vector memory access. However, this could be saved if the given offset is inside of the vector range that could make sure no IOOBE (IndexOutOfBoundaryException) happens. The masked load APIs have saved this kind of check for common cases. And this patch did the similar optimization for the masked vector store.
>>
>> The performance for the new added store masked benchmarks improves about `1.83x ~ 2.62x` on a x86 system:
>>
>> Benchmark Before After Gain Units
>> StoreMaskedBenchmark.byteStoreArrayMask 12757.936 23291.118 1.826 ops/ms
>> StoreMaskedBenchmark.doubleStoreArrayMask 1520.932 3921.616 2.578 ops/ms
>> StoreMaskedBenchmark.floatStoreArrayMask 2713.031 7122.535 2.625 ops/ms
>> StoreMaskedBenchmark.intStoreArrayMask 4113.772 8220.206 1.998 ops/ms
>> StoreMaskedBenchmark.longStoreArrayMask 1993.986 4874.148 2.444 ops/ms
>> StoreMaskedBenchmark.shortStoreArrayMask 8543.593 17821.086 2.086 ops/ms
>>
>> Similar performane gain can also be observed on ARM hardware.
>
> Xiaohong Gong has updated the pull request incrementally with one additional commit since the last revision:
>
> Wrap the offset check into a static method
Thanks for the explanation! Yeah, the main problem is Java doesn't have the direct unsigned comparison. We need the function call. From the two ways you provided, I think the second `Integer.lessThanUnsigned` looks better. But I'm not sure whether this could improve the performance a lot, although the first check `a.length - vsp.length() > 0` can be hosited out side of the loop. And this might make the codes more complex for me. Maybe we can do a pre research to find a better implementation to the unsigned comparison first. Do you think so?
-------------
PR: https://git.openjdk.java.net/jdk/pull/8620
More information about the core-libs-dev
mailing list