[GitHub] [spark] wangyum commented on a diff in pull request #37697: [SPARK-40248][SQL] Use larger number of bits to build Bloom filter
wangyum commented on code in PR #37697:
URL: https://github.com/apache/spark/pull/37697#discussion_r960736515
##
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/BloomFilterAggregate.scala:
##
@@ -55,6 +55,13 @@ case class BloomFilterAggregate(
Multiply(estimatedNumItemsExpression, Literal(8L)))
}
+ def this(child: Expression, estimatedNumItems: Long) = {
+this(child, Literal(estimatedNumItems),
+ Literal(math.min(
+BloomFilter.optimalNumOfBits(estimatedNumItems, estimatedNumItems /
30L.toDouble),
+SQLConf.get.getConf(SQLConf.RUNTIME_BLOOM_FILTER_MAX_NUM_BITS
+ }
Review Comment:
How about?
```scala
this(child, Literal(estimatedNumItems),
Literal(math.min(
BloomFilter.optimalNumOfBits(estimatedNumItems,
estimatedNumItems /
(SQLConf.get.getConf(SQLConf.RUNTIME_BLOOM_FILTER_MAX_NUM_ITEMS) /
BloomFilter.DEFAULT_FPP)),
SQLConf.get.getConf(SQLConf.RUNTIME_BLOOM_FILTER_MAX_NUM_BITS
```
The smaller `estimatedNumItems`, the smaller the `FPP`.
estimatedNumItems | FPP | numBits
-- | -- | --
RUNTIME_BLOOM_FILTER_MAX_NUM_ITEMS | DEFAULT_FPP | 29193763
200 | 0.015 | 17482271
100 | 0.0075 | 10183830
10 | 7.50E-04 | 1497636
1 | 7.50E-05 | 197688
1000 | 7.50E-06 | 24561
100 | 7.50E-07 | 2935
10 | 7.50E-08 | 341
--
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-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] wangyum commented on a diff in pull request #37697: [SPARK-40248][SQL] Use larger number of bits to build Bloom filter
wangyum commented on code in PR #37697:
URL: https://github.com/apache/spark/pull/37697#discussion_r959142316
##
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/BloomFilterAggregate.scala:
##
@@ -55,6 +55,13 @@ case class BloomFilterAggregate(
Multiply(estimatedNumItemsExpression, Literal(8L)))
}
+ def this(child: Expression, estimatedNumItems: Long) = {
+this(child, Literal(estimatedNumItems),
+ Literal(math.min(
+BloomFilter.optimalNumOfBits(estimatedNumItems, estimatedNumItems /
30L.toDouble),
Review Comment:
It can benefit all queries if `estimatedNumItems` < 9000. Especially the
number of creationSide rows is small. For example:
```scala
spark.range(1000).selectExpr("id as a", "id as b", "id as
c").write.saveAsTable("t1")
spark.range(10).selectExpr("id as x", "id as y", "id as
z").write.saveAsTable("t2")
spark.sql("set spark.sql.cbo.enabled=true")
spark.sql("ANALYZE TABLE t1 COMPUTE STATISTICS FOR COLUMNS a, b, c")
spark.sql("select * from t1 left join (select distinct * from t2) t on t1.a
= t.y where t1.b > 0").collect()
```
Before this PR | After this PR
-- | --
|
--
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-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] wangyum commented on a diff in pull request #37697: [SPARK-40248][SQL] Use larger number of bits to build Bloom filter
wangyum commented on code in PR #37697:
URL: https://github.com/apache/spark/pull/37697#discussion_r958264454
##
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/BloomFilterAggregate.scala:
##
@@ -55,6 +55,13 @@ case class BloomFilterAggregate(
Multiply(estimatedNumItemsExpression, Literal(8L)))
}
+ def this(child: Expression, estimatedNumItems: Long) = {
+this(child, Literal(estimatedNumItems),
+ Literal(math.min(
+BloomFilter.optimalNumOfBits(estimatedNumItems, estimatedNumItems /
30L.toDouble),
Review Comment:
This is used to reduce the false positive probability if `estimatedNumItems`
is small.
--
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-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
