You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2022/06/13 12:29:52 UTC
[GitHub] [spark] ulysses-you opened a new pull request, #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
ulysses-you opened a new pull request, #36856:
URL: https://github.com/apache/spark/pull/36856
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines: https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR: https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a faster review.
7. If you want to add a new configuration, please read the guideline first for naming configurations in
'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the guideline first in
'core/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
<!--
Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below.
1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers.
2. If you fix some SQL features, you can provide some references of other DBMSes.
3. If there is design documentation, please add the link.
4. If there is a discussion in the mailing list, please add the link.
-->
Pull out data type matching from `eval`. So we do not need to match data type for every row. Here is the change list:
- UnaryMinus
- Add
- Subtract
- Multiply
- BitwiseCount
- Floor
- Ceil
- RoundBase
- Hex
### Why are the changes needed?
<!--
Please clarify why the changes are needed. For instance,
1. If you propose a new API, clarify the use case for a new API.
2. If you fix a bug, you can clarify why it is a bug.
-->
Some expressions do data type matching inside `eval` but it is not friendly for performance. Which is a overhead for every execution per row.
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such as the documentation fix.
If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description and/or an example to show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change compared to the released Spark versions or within the unreleased branches such as master.
If no, write 'No'.
-->
no, only improve performance
### How was this patch tested?
<!--
If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible.
If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future.
If tests were not added, please describe why they were not added and/or why it was difficult to add.
If benchmark tests were added, please run the benchmarks in GitHub Actions for the consistent environment, and the instructions could accord to: https://spark.apache.org/developer-tools.html#github-workflow-benchmarks.
-->
it's only performance, so just pass CI
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] srowen commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
srowen commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896325472
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
I mean, why declare it this way instead of a function? I don't see what difference it makes. There is nothing expensive about creating it, so why lazy?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
ulysses-you commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896333392
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
Do you mean rewrite it to ?
```scala
private def unaryMinusFunc: Any => Any = dataType match {
......
```
The reason I pull out and make it as lazy val is: the data type is known before do eval in an expression. Let's say if the data type is integer,
- then the function can be elimiated to `input => numeric.negate(input)` during execution
- if you declare it as a function, then the function would be during execution:
```scala
dataType match {
....
case _ => input => numeric.negate(input)
```
The overhead if not about creating a function, but the data type matching elimination. So I think the lazy val is more efficient ?
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] LuciferYang commented on pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
LuciferYang commented on PR #36856:
URL: https://github.com/apache/spark/pull/36856#issuecomment-1154677592
Are there any other similar cases?
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] github-actions[bot] closed pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
github-actions[bot] closed pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
URL: https://github.com/apache/spark/pull/36856
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] github-actions[bot] commented on pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
github-actions[bot] commented on PR #36856:
URL: https://github.com/apache/spark/pull/36856#issuecomment-1255683688
We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable.
If you'd like to revive this PR, please reopen it and ask a committer to remove the Stale tag!
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
ulysses-you commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896301945
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
hmm, this would be called inside `eval` so a cached function is expected to avoid invoke by every row
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] srowen commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
srowen commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896613555
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
What is the performance improvement like? I think we'd normally ask for a benchmark. It's not a big change but not trivial; if it's a good win maybe it's worthwhile doing for more. Any numbers?
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] srowen commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
srowen commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r895807892
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
Why does this need to be a lazy val instead of just a def? I'm not sure why the line above is lazy either
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
ulysses-you commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896333392
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
Do you mean rewrite it to ?
```scala
private def unaryMinusFunc: Any => Any = dataType match {
......
```
The reason I pull out and make it as lazy val is: the data type is known before do eval in an expression. Let's say if the data type is integer,
- then the function can be elimiated to `input => numeric.negate(input)` during execution
- if you declare it as a function, then the function would be during execution:
```scala
dataType match {
....
case _ => input => numeric.negate(input)
```
The overhead is not about creating a function, but the data type matching elimination. So I think the lazy val is more efficient ?
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
ulysses-you commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896371376
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
@srowen @LuciferYang
This issue mostly happens in arithmetic like expression, and I also tried to collect other related expressions. Hope I'm not missing someone. We can still keep in mind for catching and reviewing the added expression or the new supported data type which may introduce this issue.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] srowen commented on a diff in pull request #36856: [SPARK-39455][SQL] Improve expression non-codegen code path performance by cache data type matching
Posted by GitBox <gi...@apache.org>.
srowen commented on code in PR #36856:
URL: https://github.com/apache/spark/pull/36856#discussion_r896341536
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/arithmetic.scala:
##########
@@ -53,6 +53,17 @@ case class UnaryMinus(
override def toString: String = s"-$child"
private lazy val numeric = TypeUtils.getNumeric(dataType, failOnError)
+ private lazy val unaryMinusFunc: Any => Any = dataType match {
Review Comment:
Oh I see now it depends on dataType. I understand what this does now.
How many other places would you have to change though to cover all functions? or are these the important ones?
--
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: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org