ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity
spark-instrumented-optimizer/sql/core/benchmarks/ExtractBenchmark-jdk11-results.txt

105 lines
11 KiB
Plaintext
Raw Normal View History

[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to timestamp 301 311 8 33.2 30.1 1.0X
YEAR of timestamp 930 1003 68 10.8 93.0 0.3X
YEAROFWEEK of timestamp 1019 1091 63 9.8 101.9 0.3X
QUARTER of timestamp 1151 1228 67 8.7 115.1 0.3X
MONTH of timestamp 987 1033 42 10.1 98.7 0.3X
WEEK of timestamp 1372 1449 67 7.3 137.2 0.2X
DAY of timestamp 1015 1040 22 9.8 101.5 0.3X
DAYOFWEEK of timestamp 1155 1223 59 8.7 115.5 0.3X
DOW of timestamp 1090 1170 82 9.2 109.0 0.3X
DOW_ISO of timestamp 1059 1143 93 9.4 105.9 0.3X
DAYOFWEEK_ISO of timestamp 1060 1118 51 9.4 106.0 0.3X
DOY of timestamp 958 1050 81 10.4 95.8 0.3X
HOUR of timestamp 803 817 25 12.5 80.3 0.4X
MINUTE of timestamp 707 742 30 14.1 70.7 0.4X
SECOND of timestamp 901 934 44 11.1 90.1 0.3X
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-30409][SPARK-29173][SQL][TESTS] Use `NoOp` datasource in SQL benchmarks ### What changes were proposed in this pull request? In the PR, I propose to replace `.collect()`, `.count()` and `.foreach(_ => ())` in SQL benchmarks and use the `NoOp` datasource. I added an implicit class to `SqlBasedBenchmark` with the `.noop()` method. It can be used in benchmark like: `ds.noop()`. The last one is unfolded to `ds.write.format("noop").mode(Overwrite).save()`. ### Why are the changes needed? To avoid additional overhead that `collect()` (and other actions) has. For example, `.collect()` has to convert values according to external types and pull data to the driver. This can hide actual performance regressions or improvements of benchmarked operations. ### Does this PR introduce any user-facing change? No ### How was this patch tested? Re-run all modified benchmarks using Amazon EC2. | Item | Description | | ---- | ----| | Region | us-west-2 (Oregon) | | Instance | r3.xlarge (spot instance) | | AMI | ami-06f2f779464715dc5 (ubuntu/images/hvm-ssd/ubuntu-bionic-18.04-amd64-server-20190722.1) | | Java | OpenJDK8/10 | - Run `TPCDSQueryBenchmark` using instructions from the PR #26049 ``` # `spark-tpcds-datagen` needs this. (JDK8) $ git clone https://github.com/apache/spark.git -b branch-2.4 --depth 1 spark-2.4 $ export SPARK_HOME=$PWD $ ./build/mvn clean package -DskipTests # Generate data. (JDK8) $ git clone gitgithub.com:maropu/spark-tpcds-datagen.git $ cd spark-tpcds-datagen/ $ build/mvn clean package $ mkdir -p /data/tpcds $ ./bin/dsdgen --output-location /data/tpcds/s1 // This need `Spark 2.4` ``` - Other benchmarks ran by the script: ``` #!/usr/bin/env python3 import os from sparktestsupport.shellutils import run_cmd benchmarks = [ ['sql/test', 'org.apache.spark.sql.execution.benchmark.AggregateBenchmark'], ['avro/test', 'org.apache.spark.sql.execution.benchmark.AvroReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.BloomFilterBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DataSourceReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ExtractBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.FilterPushdownBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.InExpressionBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.IntervalBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.JoinBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MakeDateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MiscBenchmark'], ['hive/test', 'org.apache.spark.sql.execution.benchmark.ObjectHashAggregateExecBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcV2NestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ParquetNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.RangeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.UDFBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideTableBenchmark'], ['hive/test', 'org.apache.spark.sql.hive.orc.OrcReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.csv.CSVBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.json.JsonBenchmark'] ] print('Set SPARK_GENERATE_BENCHMARK_FILES=1') os.environ['SPARK_GENERATE_BENCHMARK_FILES'] = '1' for b in benchmarks: print("Run benchmark: %s" % b[1]) run_cmd(['build/sbt', '%s:runMain %s' % (b[0], b[1])]) ``` Closes #27078 from MaxGekk/noop-in-benchmarks. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Maxim Gekk <maxim.gekk@databricks.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2020-01-12 16:18:19 -05:00
Invoke date_part for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to timestamp 345 349 5 29.0 34.5 1.0X
YEAR of timestamp 1050 1083 39 9.5 105.0 0.3X
YEAROFWEEK of timestamp 1125 1141 14 8.9 112.5 0.3X
QUARTER of timestamp 1103 1218 132 9.1 110.3 0.3X
MONTH of timestamp 992 1008 28 10.1 99.2 0.3X
WEEK of timestamp 1334 1403 101 7.5 133.4 0.3X
DAY of timestamp 1001 1059 52 10.0 100.1 0.3X
DAYOFWEEK of timestamp 1094 1177 77 9.1 109.4 0.3X
DOW of timestamp 1052 1128 67 9.5 105.2 0.3X
DOW_ISO of timestamp 1124 1165 39 8.9 112.4 0.3X
DAYOFWEEK_ISO of timestamp 1015 1125 101 9.9 101.5 0.3X
DOY of timestamp 1027 1080 49 9.7 102.7 0.3X
HOUR of timestamp 779 818 34 12.8 77.9 0.4X
MINUTE of timestamp 757 770 19 13.2 75.7 0.5X
SECOND of timestamp 862 898 33 11.6 86.2 0.4X
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-30409][SPARK-29173][SQL][TESTS] Use `NoOp` datasource in SQL benchmarks ### What changes were proposed in this pull request? In the PR, I propose to replace `.collect()`, `.count()` and `.foreach(_ => ())` in SQL benchmarks and use the `NoOp` datasource. I added an implicit class to `SqlBasedBenchmark` with the `.noop()` method. It can be used in benchmark like: `ds.noop()`. The last one is unfolded to `ds.write.format("noop").mode(Overwrite).save()`. ### Why are the changes needed? To avoid additional overhead that `collect()` (and other actions) has. For example, `.collect()` has to convert values according to external types and pull data to the driver. This can hide actual performance regressions or improvements of benchmarked operations. ### Does this PR introduce any user-facing change? No ### How was this patch tested? Re-run all modified benchmarks using Amazon EC2. | Item | Description | | ---- | ----| | Region | us-west-2 (Oregon) | | Instance | r3.xlarge (spot instance) | | AMI | ami-06f2f779464715dc5 (ubuntu/images/hvm-ssd/ubuntu-bionic-18.04-amd64-server-20190722.1) | | Java | OpenJDK8/10 | - Run `TPCDSQueryBenchmark` using instructions from the PR #26049 ``` # `spark-tpcds-datagen` needs this. (JDK8) $ git clone https://github.com/apache/spark.git -b branch-2.4 --depth 1 spark-2.4 $ export SPARK_HOME=$PWD $ ./build/mvn clean package -DskipTests # Generate data. (JDK8) $ git clone gitgithub.com:maropu/spark-tpcds-datagen.git $ cd spark-tpcds-datagen/ $ build/mvn clean package $ mkdir -p /data/tpcds $ ./bin/dsdgen --output-location /data/tpcds/s1 // This need `Spark 2.4` ``` - Other benchmarks ran by the script: ``` #!/usr/bin/env python3 import os from sparktestsupport.shellutils import run_cmd benchmarks = [ ['sql/test', 'org.apache.spark.sql.execution.benchmark.AggregateBenchmark'], ['avro/test', 'org.apache.spark.sql.execution.benchmark.AvroReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.BloomFilterBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DataSourceReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ExtractBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.FilterPushdownBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.InExpressionBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.IntervalBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.JoinBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MakeDateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MiscBenchmark'], ['hive/test', 'org.apache.spark.sql.execution.benchmark.ObjectHashAggregateExecBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcV2NestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ParquetNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.RangeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.UDFBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideTableBenchmark'], ['hive/test', 'org.apache.spark.sql.hive.orc.OrcReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.csv.CSVBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.json.JsonBenchmark'] ] print('Set SPARK_GENERATE_BENCHMARK_FILES=1') os.environ['SPARK_GENERATE_BENCHMARK_FILES'] = '1' for b in benchmarks: print("Run benchmark: %s" % b[1]) run_cmd(['build/sbt', '%s:runMain %s' % (b[0], b[1])]) ``` Closes #27078 from MaxGekk/noop-in-benchmarks. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Maxim Gekk <maxim.gekk@databricks.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2020-01-12 16:18:19 -05:00
Invoke extract for date: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to date 885 922 35 11.3 88.5 1.0X
YEAR of date 1100 1103 3 9.1 110.0 0.8X
YEAROFWEEK of date 1030 1076 56 9.7 103.0 0.9X
QUARTER of date 1065 1160 82 9.4 106.5 0.8X
MONTH of date 937 989 78 10.7 93.7 0.9X
WEEK of date 1389 1479 79 7.2 138.9 0.6X
DAY of date 927 997 82 10.8 92.7 1.0X
DAYOFWEEK of date 1111 1199 76 9.0 111.1 0.8X
DOW of date 1041 1149 94 9.6 104.1 0.8X
DOW_ISO of date 1049 1076 34 9.5 104.9 0.8X
DAYOFWEEK_ISO of date 1084 1134 62 9.2 108.4 0.8X
DOY of date 1042 1079 43 9.6 104.2 0.8X
HOUR of date 1792 1829 32 5.6 179.2 0.5X
MINUTE of date 1773 1827 60 5.6 177.3 0.5X
SECOND of date 2066 2085 18 4.8 206.6 0.4X
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
Invoke date_part for date: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to date 907 932 23 11.0 90.7 1.0X
YEAR of date 996 1041 55 10.0 99.6 0.9X
YEAROFWEEK of date 1032 1087 81 9.7 103.2 0.9X
QUARTER of date 1143 1193 82 8.7 114.3 0.8X
MONTH of date 953 1014 58 10.5 95.3 1.0X
WEEK of date 1260 1328 102 7.9 126.0 0.7X
DAY of date 907 927 18 11.0 90.7 1.0X
DAYOFWEEK of date 1071 1094 22 9.3 107.1 0.8X
DOW of date 1048 1077 40 9.5 104.8 0.9X
DOW_ISO of date 1024 1049 26 9.8 102.4 0.9X
DAYOFWEEK_ISO of date 1072 1082 16 9.3 107.2 0.8X
DOY of date 968 973 5 10.3 96.8 0.9X
HOUR of date 1628 1663 42 6.1 162.8 0.6X
MINUTE of date 1542 1620 69 6.5 154.2 0.6X
SECOND of date 1740 1769 33 5.7 174.0 0.5X
[SPARK-31119][SQL] Add interval value support for extract expression as extract source ### What changes were proposed in this pull request? ``` <extract expression> ::= EXTRACT <left paren> <extract field> FROM <extract source> <right paren> <extract source> ::= <datetime value expression> | <interval value expression> ``` We now only support datetime values as extract source for `extract` expression but it's alternative function `date_part` supports both datetime and interval. This pr adds interval value support for `extract` expression as extract source ### Why are the changes needed? For ANSI compliance and the semantic consistency between extract and `date_part`, we support intervals for extract expressions. ### Does this PR introduce any user-facing change? yes, in the `extract(abc from xyz)` expression, the `xyz` can be intervals ### How was this patch tested? add unit tests Closes #27876 from yaooqinn/SPARK-31119. Authored-by: Kent Yao <yaooqinn@hotmail.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>
2020-03-18 00:29:39 -04:00
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-31119][SQL] Add interval value support for extract expression as extract source ### What changes were proposed in this pull request? ``` <extract expression> ::= EXTRACT <left paren> <extract field> FROM <extract source> <right paren> <extract source> ::= <datetime value expression> | <interval value expression> ``` We now only support datetime values as extract source for `extract` expression but it's alternative function `date_part` supports both datetime and interval. This pr adds interval value support for `extract` expression as extract source ### Why are the changes needed? For ANSI compliance and the semantic consistency between extract and `date_part`, we support intervals for extract expressions. ### Does this PR introduce any user-facing change? yes, in the `extract(abc from xyz)` expression, the `xyz` can be intervals ### How was this patch tested? add unit tests Closes #27876 from yaooqinn/SPARK-31119. Authored-by: Kent Yao <yaooqinn@hotmail.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>
2020-03-18 00:29:39 -04:00
Invoke extract for interval: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to interval 1413 1431 21 7.1 141.3 1.0X
YEAR of interval 1386 1424 34 7.2 138.6 1.0X
MONTH of interval 1419 1434 17 7.0 141.9 1.0X
DAY of interval 1411 1428 15 7.1 141.1 1.0X
HOUR of interval 1444 1461 24 6.9 144.4 1.0X
MINUTE of interval 1435 1463 27 7.0 143.5 1.0X
SECOND of interval 1537 1565 30 6.5 153.7 0.9X
[SPARK-30409][SPARK-29173][SQL][TESTS] Use `NoOp` datasource in SQL benchmarks ### What changes were proposed in this pull request? In the PR, I propose to replace `.collect()`, `.count()` and `.foreach(_ => ())` in SQL benchmarks and use the `NoOp` datasource. I added an implicit class to `SqlBasedBenchmark` with the `.noop()` method. It can be used in benchmark like: `ds.noop()`. The last one is unfolded to `ds.write.format("noop").mode(Overwrite).save()`. ### Why are the changes needed? To avoid additional overhead that `collect()` (and other actions) has. For example, `.collect()` has to convert values according to external types and pull data to the driver. This can hide actual performance regressions or improvements of benchmarked operations. ### Does this PR introduce any user-facing change? No ### How was this patch tested? Re-run all modified benchmarks using Amazon EC2. | Item | Description | | ---- | ----| | Region | us-west-2 (Oregon) | | Instance | r3.xlarge (spot instance) | | AMI | ami-06f2f779464715dc5 (ubuntu/images/hvm-ssd/ubuntu-bionic-18.04-amd64-server-20190722.1) | | Java | OpenJDK8/10 | - Run `TPCDSQueryBenchmark` using instructions from the PR #26049 ``` # `spark-tpcds-datagen` needs this. (JDK8) $ git clone https://github.com/apache/spark.git -b branch-2.4 --depth 1 spark-2.4 $ export SPARK_HOME=$PWD $ ./build/mvn clean package -DskipTests # Generate data. (JDK8) $ git clone gitgithub.com:maropu/spark-tpcds-datagen.git $ cd spark-tpcds-datagen/ $ build/mvn clean package $ mkdir -p /data/tpcds $ ./bin/dsdgen --output-location /data/tpcds/s1 // This need `Spark 2.4` ``` - Other benchmarks ran by the script: ``` #!/usr/bin/env python3 import os from sparktestsupport.shellutils import run_cmd benchmarks = [ ['sql/test', 'org.apache.spark.sql.execution.benchmark.AggregateBenchmark'], ['avro/test', 'org.apache.spark.sql.execution.benchmark.AvroReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.BloomFilterBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DataSourceReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ExtractBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.FilterPushdownBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.InExpressionBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.IntervalBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.JoinBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MakeDateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MiscBenchmark'], ['hive/test', 'org.apache.spark.sql.execution.benchmark.ObjectHashAggregateExecBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcV2NestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ParquetNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.RangeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.UDFBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideTableBenchmark'], ['hive/test', 'org.apache.spark.sql.hive.orc.OrcReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.csv.CSVBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.json.JsonBenchmark'] ] print('Set SPARK_GENERATE_BENCHMARK_FILES=1') os.environ['SPARK_GENERATE_BENCHMARK_FILES'] = '1' for b in benchmarks: print("Run benchmark: %s" % b[1]) run_cmd(['build/sbt', '%s:runMain %s' % (b[0], b[1])]) ``` Closes #27078 from MaxGekk/noop-in-benchmarks. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Maxim Gekk <maxim.gekk@databricks.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2020-01-12 16:18:19 -05:00
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
OpenJDK 64-Bit Server VM 11.0.10+9-LTS on Linux 5.4.0-1043-azure
Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
[SPARK-30409][SPARK-29173][SQL][TESTS] Use `NoOp` datasource in SQL benchmarks ### What changes were proposed in this pull request? In the PR, I propose to replace `.collect()`, `.count()` and `.foreach(_ => ())` in SQL benchmarks and use the `NoOp` datasource. I added an implicit class to `SqlBasedBenchmark` with the `.noop()` method. It can be used in benchmark like: `ds.noop()`. The last one is unfolded to `ds.write.format("noop").mode(Overwrite).save()`. ### Why are the changes needed? To avoid additional overhead that `collect()` (and other actions) has. For example, `.collect()` has to convert values according to external types and pull data to the driver. This can hide actual performance regressions or improvements of benchmarked operations. ### Does this PR introduce any user-facing change? No ### How was this patch tested? Re-run all modified benchmarks using Amazon EC2. | Item | Description | | ---- | ----| | Region | us-west-2 (Oregon) | | Instance | r3.xlarge (spot instance) | | AMI | ami-06f2f779464715dc5 (ubuntu/images/hvm-ssd/ubuntu-bionic-18.04-amd64-server-20190722.1) | | Java | OpenJDK8/10 | - Run `TPCDSQueryBenchmark` using instructions from the PR #26049 ``` # `spark-tpcds-datagen` needs this. (JDK8) $ git clone https://github.com/apache/spark.git -b branch-2.4 --depth 1 spark-2.4 $ export SPARK_HOME=$PWD $ ./build/mvn clean package -DskipTests # Generate data. (JDK8) $ git clone gitgithub.com:maropu/spark-tpcds-datagen.git $ cd spark-tpcds-datagen/ $ build/mvn clean package $ mkdir -p /data/tpcds $ ./bin/dsdgen --output-location /data/tpcds/s1 // This need `Spark 2.4` ``` - Other benchmarks ran by the script: ``` #!/usr/bin/env python3 import os from sparktestsupport.shellutils import run_cmd benchmarks = [ ['sql/test', 'org.apache.spark.sql.execution.benchmark.AggregateBenchmark'], ['avro/test', 'org.apache.spark.sql.execution.benchmark.AvroReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.BloomFilterBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DataSourceReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.DateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ExtractBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.FilterPushdownBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.InExpressionBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.IntervalBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.JoinBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MakeDateTimeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.MiscBenchmark'], ['hive/test', 'org.apache.spark.sql.execution.benchmark.ObjectHashAggregateExecBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.OrcV2NestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.ParquetNestedSchemaPruningBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.RangeBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.UDFBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.benchmark.WideTableBenchmark'], ['hive/test', 'org.apache.spark.sql.hive.orc.OrcReadBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.csv.CSVBenchmark'], ['sql/test', 'org.apache.spark.sql.execution.datasources.json.JsonBenchmark'] ] print('Set SPARK_GENERATE_BENCHMARK_FILES=1') os.environ['SPARK_GENERATE_BENCHMARK_FILES'] = '1' for b in benchmarks: print("Run benchmark: %s" % b[1]) run_cmd(['build/sbt', '%s:runMain %s' % (b[0], b[1])]) ``` Closes #27078 from MaxGekk/noop-in-benchmarks. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Maxim Gekk <maxim.gekk@databricks.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2020-01-12 16:18:19 -05:00
Invoke date_part for interval: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------------------------------
[SPARK-34950][TESTS] Update benchmark results to the ones created by GitHub Actions machines ### What changes were proposed in this pull request? https://github.com/apache/spark/pull/32015 added a way to run benchmarks much more easily in the same GitHub Actions build. This PR updates the benchmark results by using the way. **NOTE** that looks like GitHub Actions use four types of CPU given my observations: - Intel(R) Xeon(R) Platinum 8171M CPU 2.60GHz - Intel(R) Xeon(R) CPU E5-2673 v4 2.30GHz - Intel(R) Xeon(R) CPU E5-2673 v3 2.40GHz - Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz Given my quick research, seems like they perform roughly similarly: ![Screen Shot 2021-04-03 at 9 31 23 PM](https://user-images.githubusercontent.com/6477701/113478478-f4b57b80-94c3-11eb-9047-f81ca8c59672.png) I couldn't find enough information about Intel(R) Xeon(R) Platinum 8272CL CPU 2.60GHz but the performance seems roughly similar given the numbers. So shouldn't be a big deal especially given that this way is much easier, encourages contributors to run more and guarantee the same number of cores and same memory with the same softwares. ### Why are the changes needed? To have a base line of the benchmarks accordingly. ### Does this PR introduce _any_ user-facing change? No, dev-only. ### How was this patch tested? It was generated from: - [Run benchmarks: * (JDK 11)](https://github.com/HyukjinKwon/spark/actions/runs/713575465) - [Run benchmarks: * (JDK 8)](https://github.com/HyukjinKwon/spark/actions/runs/713154337) Closes #32044 from HyukjinKwon/SPARK-34950. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Max Gekk <max.gekk@gmail.com>
2021-04-03 16:02:56 -04:00
cast to interval 1434 1450 15 7.0 143.4 1.0X
YEAR of interval 1423 1443 18 7.0 142.3 1.0X
MONTH of interval 1432 1452 22 7.0 143.2 1.0X
DAY of interval 1451 1467 14 6.9 145.1 1.0X
HOUR of interval 1415 1431 23 7.1 141.5 1.0X
MINUTE of interval 1433 1458 36 7.0 143.3 1.0X
SECOND of interval 1543 1556 13 6.5 154.3 0.9X
[SPARK-29190][SQL] Optimize `extract`/`date_part` for the milliseconds `field` ### What changes were proposed in this pull request? Changed the `DateTimeUtils.getMilliseconds()` by avoiding the decimal division, and replacing it by setting scale and precision while converting microseconds to the decimal type. ### Why are the changes needed? This improves performance of `extract` and `date_part()` by more than **50 times**: Before: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 397 428 45 25.2 39.7 1.0X MILLISECONDS of timestamp 36723 36761 63 0.3 3672.3 0.0X ``` After: ``` Invoke extract for timestamp: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cast to timestamp 278 284 6 36.0 27.8 1.0X MILLISECONDS of timestamp 592 606 13 16.9 59.2 0.5X ``` ### Does this PR introduce any user-facing change? No ### How was this patch tested? By existing test suite - `DateExpressionsSuite` Closes #25871 from MaxGekk/optimize-epoch-millis. Lead-authored-by: Maxim Gekk <max.gekk@gmail.com> Co-authored-by: Dongjoon Hyun <dhyun@apple.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
2019-09-22 00:11:31 -04:00
Reference in a new issue Copy permalink