On Wed, 11 Oct 2023 17:28:12 GMT, Srinivas Vamsi Parasa <d...@openjdk.org> wrote:
>> The goal of this PR is to address the follow-up comments to the SIMD >> accelerated sort PR (#14227) which implemented AVX512 intrinsics for >> Arrays.sort() methods. >> The proposed changes are: >> >> 1) Restriction of the AVX512 sort acceleration to only Intel CPUs. A >> performance regression (due to micro-architectural differences) was reported >> for AMD Zen4 CPUs in the comments section of PR. >> 2) Addressing the build failure due to a bug in GCC 12 (which was fixed in >> version 12.3.1). The details of the bug are at: >> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105593 >> 3) Minor changes in Javadoc strings > > Srinivas Vamsi Parasa has updated the pull request incrementally with one > additional commit since the last revision: > > Add @ForceInline annotation to insertion and mixedInsertion sort > > > Also @forceinline in these changes only works for case when new > > > intrinsics are not used. I would suggest to adapt/update JMH benchmark to > > > cover all cases and see effect @forceinline without intrinsics. That will > > > tell us which @forceinline annotations are needed. > > > > > > Added `@ForceInline` annotations to `insertionSort` and > > `mixedInsertionSort` as it is helping arrays of small sizes when intrinsics > > are disabled. > > Thanks, Vamsi > > @vamsi-parasa Please revert changes (adding @forceinline to insertionSort and > mixedInsertionSort) - I checked: initinal version works faster. @iaroslavski Vladimir, have you used `-XX:+UnlockDiagnosticVMOptions -XX:DisableIntrinsic=_arraySort,_arrayPartition` to disable the intrinsics? Also, it helps to have one warmup iteration of 30 secs to reduce run-to-run variance. ------------- PR Comment: https://git.openjdk.org/jdk/pull/16124#issuecomment-1758437209