dongjoon-hyun commented on code in PR #47743: URL: https://github.com/apache/spark/pull/47743#discussion_r1715646176
########## sql/core/benchmarks/DataSourceReadBenchmark-jdk21-results.txt: ########## @@ -2,437 +2,437 @@ SQL Single Numeric Column Scan ================================================================================================ -OpenJDK 64-Bit Server VM 21.0.3+9-LTS on Linux 6.5.0-1023-azure +OpenJDK 64-Bit Server VM 21.0.4+7-LTS on Linux 6.5.0-1025-azure AMD EPYC 7763 64-Core Processor SQL Single BOOLEAN Column Scan: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ -SQL CSV 9728 9736 11 1.6 618.5 1.0X -SQL Json 7883 7959 107 2.0 501.2 1.2X -SQL Json with UnsafeRow 8410 8425 21 1.9 534.7 1.2X -SQL Parquet Vectorized: DataPageV1 80 94 7 195.5 5.1 120.9X -SQL Parquet Vectorized: DataPageV2 93 107 8 169.1 5.9 104.6X -SQL Parquet MR: DataPageV1 1767 1774 9 8.9 112.4 5.5X -SQL Parquet MR: DataPageV2 1650 1651 1 9.5 104.9 5.9X -SQL ORC Vectorized 120 131 9 131.2 7.6 81.1X -SQL ORC MR 1503 1523 28 10.5 95.6 6.5X - -OpenJDK 64-Bit Server VM 21.0.3+9-LTS on Linux 6.5.0-1023-azure +SQL CSV 10214 10246 45 1.5 649.4 1.0X +SQL Json 7831 7865 48 2.0 497.9 1.3X +SQL Json with UnsafeRow 8565 8571 8 1.8 544.6 1.2X +SQL Parquet Vectorized: DataPageV1 81 96 11 193.3 5.2 125.6X +SQL Parquet Vectorized: DataPageV2 201 210 8 78.4 12.8 50.9X Review Comment: `DataPageV2` becomes very slower than `DataPageV1` here and next benchmark. Please note that the previous result was generated when we upgraded to `Apache Parquet 1.14.1`. - #46447 So, if there is a reason for this, it's not Apache Parquet dependency. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
