On Mon, 27 Apr 2026 09:45:22 GMT, Mohamed Issa <[email protected]> wrote:
>> Added float/double benchmarks to `MinMaxVector`. >> >> Here are some sample results: >> >> darwin/aarch64 (m4 pro) >> >> Benchmark (probability) (range) (seed) >> (size) Mode Cnt Score Error Units >> MinMaxVector.doubleClippingRange N/A 90 0 >> 1000 thrpt 4 14780.471 ± 96.860 ops/ms >> MinMaxVector.doubleClippingRange N/A 100 0 >> 1000 thrpt 4 14712.065 ± 94.435 ops/ms >> MinMaxVector.doubleLoopMax 50 N/A N/A >> 2048 thrpt 4 5068.118 ± 15.706 ops/ms >> MinMaxVector.doubleLoopMax 80 N/A N/A >> 2048 thrpt 4 5077.801 ± 28.764 ops/ms >> MinMaxVector.doubleLoopMax 100 N/A N/A >> 2048 thrpt 4 5069.538 ± 28.218 ops/ms >> MinMaxVector.doubleLoopMin 50 N/A N/A >> 2048 thrpt 4 5051.947 ± 27.410 ops/ms >> MinMaxVector.doubleLoopMin 80 N/A N/A >> 2048 thrpt 4 4955.209 ± 36.664 ops/ms >> MinMaxVector.doubleLoopMin 100 N/A N/A >> 2048 thrpt 4 4926.185 ± 34.539 ops/ms >> MinMaxVector.doubleReductionMultiplyMax 50 N/A N/A >> 2048 thrpt 4 2403.890 ± 21.603 ops/ms >> MinMaxVector.doubleReductionMultiplyMax 80 N/A N/A >> 2048 thrpt 4 2403.574 ± 25.091 ops/ms >> MinMaxVector.doubleReductionMultiplyMax 100 N/A N/A >> 2048 thrpt 4 2403.080 ± 4.945 ops/ms >> MinMaxVector.doubleReductionMultiplyMin 50 N/A N/A >> 2048 thrpt 4 2403.787 ± 3.896 ops/ms >> MinMaxVector.doubleReductionMultiplyMin 80 N/A N/A >> 2048 thrpt 4 2402.399 ± 13.312 ops/ms >> MinMaxVector.doubleReductionMultiplyMin 100 N/A N/A >> 2048 thrpt 4 2402.061 ± 21.479 ops/ms >> MinMaxVector.doubleReductionSimpleMax 50 N/A N/A >> 2048 thrpt 4 2531.053 ± 6.344 ops/ms >> MinMaxVector.doubleReductionSimpleMax 80 N/A N/A >> 2048 thrpt 4 2526.707 ± 41.029 ops/ms >> MinMaxVector.doubleReductionSimpleMax 100 N/A N/A >> 2048 thrpt 4 2514.099 ± 9.836 ops/ms >> MinMaxVector.doubleReductionSimpleMin 50 N/A N/A >> 2048 thrpt 4 2502.677 ± 5.173 ops/ms >> MinMaxVector.doubleReductionSimpleMin ... > > Could you post the Xeon SKU you used? @missa-prime can you re-review this? ------------- PR Comment: https://git.openjdk.org/jdk/pull/30844#issuecomment-5396266462
