cb0cddffe9
## What changes were proposed in this pull request? This pr proposed to split aggregation code into small functions in `HashAggregateExec`. In #18810, we got performance regression if JVMs didn't compile too long functions. I checked and I found the codegen of `HashAggregateExec` frequently goes over the limit when a query has too many aggregate functions (e.g., q66 in TPCDS). The current master places all the generated aggregation code in a single function. In this pr, I modified the code to assign an individual function for each aggregate function (e.g., `SUM` and `AVG`). For example, in a query `SELECT SUM(a), AVG(a) FROM VALUES(1) t(a)`, the proposed code defines two functions for `SUM(a)` and `AVG(a)` as follows; - generated code with this pr (https://gist.github.com/maropu/812990012bc967a78364be0fa793f559): ``` /* 173 */ private void agg_doConsume_0(InternalRow inputadapter_row_0, long agg_expr_0_0, boolean agg_exprIsNull_0_0, double agg_expr_1_0, boolean agg_exprIsNull_1_0, long agg_expr_2_0, boolean agg_exprIsNull_2_0) throws java.io.IOException { /* 174 */ // do aggregate /* 175 */ // common sub-expressions /* 176 */ /* 177 */ // evaluate aggregate functions and update aggregation buffers /* 178 */ agg_doAggregate_sum_0(agg_exprIsNull_0_0, agg_expr_0_0); /* 179 */ agg_doAggregate_avg_0(agg_expr_1_0, agg_exprIsNull_1_0, agg_exprIsNull_2_0, agg_expr_2_0); /* 180 */ /* 181 */ } ... /* 071 */ private void agg_doAggregate_avg_0(double agg_expr_1_0, boolean agg_exprIsNull_1_0, boolean agg_exprIsNull_2_0, long agg_expr_2_0) throws java.io.IOException { /* 072 */ // do aggregate for avg /* 073 */ // evaluate aggregate function /* 074 */ boolean agg_isNull_19 = true; /* 075 */ double agg_value_19 = -1.0; ... /* 114 */ private void agg_doAggregate_sum_0(boolean agg_exprIsNull_0_0, long agg_expr_0_0) throws java.io.IOException { /* 115 */ // do aggregate for sum /* 116 */ // evaluate aggregate function /* 117 */ agg_agg_isNull_11_0 = true; /* 118 */ long agg_value_11 = -1L; ``` - generated code in the current master (https://gist.github.com/maropu/e9d772af2c98d8991a6a5f0af7841760) ``` /* 059 */ private void agg_doConsume_0(InternalRow localtablescan_row_0, int agg_expr_0_0) throws java.io.IOException { /* 060 */ // do aggregate /* 061 */ // common sub-expressions /* 062 */ boolean agg_isNull_4 = false; /* 063 */ long agg_value_4 = -1L; /* 064 */ if (!false) { /* 065 */ agg_value_4 = (long) agg_expr_0_0; /* 066 */ } /* 067 */ // evaluate aggregate function /* 068 */ agg_agg_isNull_7_0 = true; /* 069 */ long agg_value_7 = -1L; /* 070 */ do { /* 071 */ if (!agg_bufIsNull_0) { /* 072 */ agg_agg_isNull_7_0 = false; /* 073 */ agg_value_7 = agg_bufValue_0; /* 074 */ continue; /* 075 */ } /* 076 */ /* 077 */ boolean agg_isNull_9 = false; /* 078 */ long agg_value_9 = -1L; /* 079 */ if (!false) { /* 080 */ agg_value_9 = (long) 0; /* 081 */ } /* 082 */ if (!agg_isNull_9) { /* 083 */ agg_agg_isNull_7_0 = false; /* 084 */ agg_value_7 = agg_value_9; /* 085 */ continue; /* 086 */ } /* 087 */ /* 088 */ } while (false); /* 089 */ /* 090 */ long agg_value_6 = -1L; /* 091 */ /* 092 */ agg_value_6 = agg_value_7 + agg_value_4; /* 093 */ boolean agg_isNull_11 = true; /* 094 */ double agg_value_11 = -1.0; /* 095 */ /* 096 */ if (!agg_bufIsNull_1) { /* 097 */ agg_agg_isNull_13_0 = true; /* 098 */ double agg_value_13 = -1.0; /* 099 */ do { /* 100 */ boolean agg_isNull_14 = agg_isNull_4; /* 101 */ double agg_value_14 = -1.0; /* 102 */ if (!agg_isNull_4) { /* 103 */ agg_value_14 = (double) agg_value_4; /* 104 */ } /* 105 */ if (!agg_isNull_14) { /* 106 */ agg_agg_isNull_13_0 = false; /* 107 */ agg_value_13 = agg_value_14; /* 108 */ continue; /* 109 */ } /* 110 */ /* 111 */ boolean agg_isNull_15 = false; /* 112 */ double agg_value_15 = -1.0; /* 113 */ if (!false) { /* 114 */ agg_value_15 = (double) 0; /* 115 */ } /* 116 */ if (!agg_isNull_15) { /* 117 */ agg_agg_isNull_13_0 = false; /* 118 */ agg_value_13 = agg_value_15; /* 119 */ continue; /* 120 */ } /* 121 */ /* 122 */ } while (false); /* 123 */ /* 124 */ agg_isNull_11 = false; // resultCode could change nullability. /* 125 */ /* 126 */ agg_value_11 = agg_bufValue_1 + agg_value_13; /* 127 */ /* 128 */ } /* 129 */ boolean agg_isNull_17 = false; /* 130 */ long agg_value_17 = -1L; /* 131 */ if (!false && agg_isNull_4) { /* 132 */ agg_isNull_17 = agg_bufIsNull_2; /* 133 */ agg_value_17 = agg_bufValue_2; /* 134 */ } else { /* 135 */ boolean agg_isNull_20 = true; /* 136 */ long agg_value_20 = -1L; /* 137 */ /* 138 */ if (!agg_bufIsNull_2) { /* 139 */ agg_isNull_20 = false; // resultCode could change nullability. /* 140 */ /* 141 */ agg_value_20 = agg_bufValue_2 + 1L; /* 142 */ /* 143 */ } /* 144 */ agg_isNull_17 = agg_isNull_20; /* 145 */ agg_value_17 = agg_value_20; /* 146 */ } /* 147 */ // update aggregation buffer /* 148 */ agg_bufIsNull_0 = false; /* 149 */ agg_bufValue_0 = agg_value_6; /* 150 */ /* 151 */ agg_bufIsNull_1 = agg_isNull_11; /* 152 */ agg_bufValue_1 = agg_value_11; /* 153 */ /* 154 */ agg_bufIsNull_2 = agg_isNull_17; /* 155 */ agg_bufValue_2 = agg_value_17; /* 156 */ /* 157 */ } ``` You can check the previous discussion in https://github.com/apache/spark/pull/19082 ## How was this patch tested? Existing tests Closes #20965 from maropu/SPARK-21870-2. Authored-by: Takeshi Yamamuro <yamamuro@apache.org> Signed-off-by: Wenchen Fan <wenchen@databricks.com> |
||
|---|---|---|
| .github | ||
| assembly | ||
| bin | ||
| build | ||
| common | ||
| conf | ||
| core | ||
| data | ||
| dev | ||
| docs | ||
| examples | ||
| external | ||
| graph | ||
| graphx | ||
| hadoop-cloud | ||
| launcher | ||
| licenses | ||
| licenses-binary | ||
| mllib | ||
| mllib-local | ||
| project | ||
| python | ||
| R | ||
| repl | ||
| resource-managers | ||
| sbin | ||
| sql | ||
| streaming | ||
| tools | ||
| .gitattributes | ||
| .gitignore | ||
| appveyor.yml | ||
| CONTRIBUTING.md | ||
| LICENSE | ||
| LICENSE-binary | ||
| NOTICE | ||
| NOTICE-binary | ||
| pom.xml | ||
| README.md | ||
| scalastyle-config.xml | ||
Apache Spark
Spark is a unified analytics engine for large-scale data processing. It provides high-level APIs in Scala, Java, Python, and R, and an optimized engine that supports general computation graphs for data analysis. It also supports a rich set of higher-level tools including Spark SQL for SQL and DataFrames, MLlib for machine learning, GraphX for graph processing, and Structured Streaming for stream processing.
Online Documentation
You can find the latest Spark documentation, including a programming guide, on the project web page. This README file only contains basic setup instructions.
Building Spark
Spark is built using Apache Maven. To build Spark and its example programs, run:
./build/mvn -DskipTests clean package
(You do not need to do this if you downloaded a pre-built package.)
You can build Spark using more than one thread by using the -T option with Maven, see "Parallel builds in Maven 3". More detailed documentation is available from the project site, at "Building Spark".
For general development tips, including info on developing Spark using an IDE, see "Useful Developer Tools".
Interactive Scala Shell
The easiest way to start using Spark is through the Scala shell:
./bin/spark-shell
Try the following command, which should return 1,000,000,000:
scala> spark.range(1000 * 1000 * 1000).count()
Interactive Python Shell
Alternatively, if you prefer Python, you can use the Python shell:
./bin/pyspark
And run the following command, which should also return 1,000,000,000:
>>> spark.range(1000 * 1000 * 1000).count()
Example Programs
Spark also comes with several sample programs in the examples directory.
To run one of them, use ./bin/run-example <class> [params]. For example:
./bin/run-example SparkPi
will run the Pi example locally.
You can set the MASTER environment variable when running examples to submit
examples to a cluster. This can be a mesos:// or spark:// URL,
"yarn" to run on YARN, and "local" to run
locally with one thread, or "local[N]" to run locally with N threads. You
can also use an abbreviated class name if the class is in the examples
package. For instance:
MASTER=spark://host:7077 ./bin/run-example SparkPi
Many of the example programs print usage help if no params are given.
Running Tests
Testing first requires building Spark. Once Spark is built, tests can be run using:
./dev/run-tests
Please see the guidance on how to run tests for a module, or individual tests.
There is also a Kubernetes integration test, see resource-managers/kubernetes/integration-tests/README.md
A Note About Hadoop Versions
Spark uses the Hadoop core library to talk to HDFS and other Hadoop-supported storage systems. Because the protocols have changed in different versions of Hadoop, you must build Spark against the same version that your cluster runs.
Please refer to the build documentation at "Specifying the Hadoop Version and Enabling YARN" for detailed guidance on building for a particular distribution of Hadoop, including building for particular Hive and Hive Thriftserver distributions.
Configuration
Please refer to the Configuration Guide in the online documentation for an overview on how to configure Spark.
Contributing
Please review the Contribution to Spark guide for information on how to get started contributing to the project.