On Fri, 18 Sep 2026 04:29:14 GMT, Jatin Bhateja <[email protected]> wrote:
> Hi All, > > Currently, > [ReplHF_reg](https://github.com/openjdk/jdk/blob/master/src/hotspot/cpu/x86/x86.ad#L18646) > selection pattern > which accepts regF src operand, first copies it to a GPR using _evmovw_ > instruction (which has 3 cycles latency on x86 targets supporting AVX512-FP16 > feature) and then broadcast it using _vpbroadcastw_ instruction. > > On targets supporting AVX512BW feature this sequence can be replaced by a > single _vpbroadcastw_ instruction which can directly broadcast XMM source > thus saving costly XMM to GPR move. > > This shows 2-3% performance improvement on AVX512-FP16 systems for benchmark > included with the patch. > > > Baseline:- > Benchmark Mode Cnt Score Error Units > ReplHFRegBenchmark.manyBroadcasts thrpt 5 16.587 ± 0.041 ops/ms > Withchange:- > Benchmark Mode Cnt Score Error Units > ReplHFRegBenchmark.manyBroadcasts thrpt 5 16.978 ± 0.033 ops/ms > > > Kindly review and share your feedback. > > Best Regards, > Jatin > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). This pull request has now been integrated. Changeset: 042f64e8 Author: Jatin Bhateja <[email protected]> URL: https://git.openjdk.org/jdk/commit/042f64e81f8e11e677062717d0f5bfc43212e7ab Stats: 97 lines in 3 files changed: 90 ins; 2 del; 5 mod 8392548: Optimize instruction sequence for x86 ReplHF_reg instruction selection pattern Reviewed-by: sviswanathan, qamai ------------- PR: https://git.openjdk.org/jdk/pull/32938
