RFR: 8317976: Optimize SIMD sort for AMD Zen 4 [v2]
Srinivas Vamsi Parasa
sparasa at openjdk.org
Mon Mar 17 17:06:11 UTC 2025
On Mon, 17 Mar 2025 15:28:12 GMT, Rohit Arul Raj <rraj at openjdk.org> wrote:
>> This change enables optimized SIMD sort for AMD Zen 4 (AVX2) & Zen 5 (AVX512).
>>
>> JTREG Tests: Completed Tier1 & Tier2 tests on Zen4 & Zen5 - No Regressions.
>>
>> Attaching ArraySort performance data for Zen4 & Zen5.
>> [Zen4-ArraySort-Data.txt](https://github.com/user-attachments/files/19245831/Zen4-ArraySort-Data.txt)
>> [Zen5-ArraySort-Data.txt](https://github.com/user-attachments/files/19245833/Zen5-ArraySort-Data.txt)
>
> Rohit Arul Raj has updated the pull request incrementally with one additional commit since the last revision:
>
> create a separate method to check for cpu's supporting avx512 version of simd sort
src/hotspot/cpu/x86/stubGenerator_x86_64.cpp line 4318:
> 4316: // Load x86_64_sort library on supported hardware to enable SIMD sort and partition intrinsics
> 4317:
> 4318: if (VM_Version::supports_avx512dq() || VM_Version::supports_avx2()) {
Shouldn't you check for `VM_Version::supports_avx512_simd_sort()` here as well?
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/24053#discussion_r1999252275
More information about the hotspot-compiler-dev
mailing list