uschindler commented on PR #12706: URL: https://github.com/apache/lucene/pull/12706#issuecomment-1890925338
I did a quick benchmark of Vector API and compared Java 22-ea+31 vs. Java 21.0.1 (AMD Ryzen 7 3700X 8-Core Processor): ``` $ ~jenkins/tools/java/64bit/hotspot/jdk-22-ea+31/bin/java --module-path lucene/benchmark-jmh/build/benchmarks --module org.apache.lucene.benchmark.jmh VectorUtilBenchmark -p size=1024 Jan 14, 2024 11:12:14 AM org.apache.lucene.internal.vectorization.PanamaVectorizationProvider <init> INFO: Java vector incubator API enabled; uses preferredBitSize=256; FMA enabled Benchmark (size) Mode Cnt Score Error Units VectorUtilBenchmark.binaryCosineScalar 1024 thrpt 15 0.632 ± 0.004 ops/us VectorUtilBenchmark.binaryCosineVector 1024 thrpt 15 4.635 ± 0.049 ops/us VectorUtilBenchmark.binaryDotProductScalar 1024 thrpt 15 1.904 ± 0.030 ops/us VectorUtilBenchmark.binaryDotProductVector 1024 thrpt 15 12.080 ± 0.101 ops/us VectorUtilBenchmark.binarySquareScalar 1024 thrpt 15 1.566 ± 0.029 ops/us VectorUtilBenchmark.binarySquareVector 1024 thrpt 15 11.060 ± 0.141 ops/us VectorUtilBenchmark.floatCosineScalar 1024 thrpt 15 1.464 ± 0.019 ops/us VectorUtilBenchmark.floatCosineVector 1024 thrpt 75 10.497 ± 0.042 ops/us VectorUtilBenchmark.floatDotProductScalar 1024 thrpt 15 3.887 ± 0.036 ops/us VectorUtilBenchmark.floatDotProductVector 1024 thrpt 75 20.890 ± 0.446 ops/us VectorUtilBenchmark.floatSquareScalar 1024 thrpt 15 3.063 ± 0.034 ops/us VectorUtilBenchmark.floatSquareVector 1024 thrpt 75 20.096 ± 0.198 ops/us $ ~jenkins/tools/java/64bit/hotspot/jdk-21.0.1/bin/java --module-path lucene/benchmark-jmh/build/benchmarks --module org.apache.lucene.benchmark.jmh VectorUtilBenchmark -p size=1024 Jan 14, 2024 11:28:42 AM org.apache.lucene.internal.vectorization.PanamaVectorizationProvider <init> INFO: Java vector incubator API enabled; uses preferredBitSize=256; FMA enabled Benchmark (size) Mode Cnt Score Error Units VectorUtilBenchmark.binaryCosineScalar 1024 thrpt 15 0.627 ± 0.005 ops/us VectorUtilBenchmark.binaryCosineVector 1024 thrpt 15 4.684 ± 0.044 ops/us VectorUtilBenchmark.binaryDotProductScalar 1024 thrpt 15 1.936 ± 0.026 ops/us VectorUtilBenchmark.binaryDotProductVector 1024 thrpt 15 12.235 ± 0.117 ops/us VectorUtilBenchmark.binarySquareScalar 1024 thrpt 15 1.590 ± 0.014 ops/us VectorUtilBenchmark.binarySquareVector 1024 thrpt 15 11.114 ± 0.143 ops/us VectorUtilBenchmark.floatCosineScalar 1024 thrpt 15 1.442 ± 0.025 ops/us VectorUtilBenchmark.floatCosineVector 1024 thrpt 75 10.370 ± 0.049 ops/us VectorUtilBenchmark.floatDotProductScalar 1024 thrpt 15 3.866 ± 0.028 ops/us VectorUtilBenchmark.floatDotProductVector 1024 thrpt 75 19.050 ± 0.547 ops/us VectorUtilBenchmark.floatSquareScalar 1024 thrpt 15 3.094 ± 0.028 ops/us VectorUtilBenchmark.floatSquareVector 1024 thrpt 75 18.836 ± 0.343 ops/us ``` So all looks fine. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
