release-20.1: colexec: fix performance inefficiency in materializer #48732
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
We mistakenly were passing `sqlbase.DatumAlloc` by value, and not by pointer, and as a result we would always be allocating 16 datums but using only 1 - i.e. we were not only not pooling the allocations, but actually making a bunch of useless allocations as well. This inefficiency becomes noticeable when the vectorized query returns many rows and when we have wrapped processors and those processors get a lot of input rows - in all cases when we need to materialize a lot. For example, TPC-H query 16 sees about 10% improvement (it returns 18k rows) and TPC-DS query 6 sees 2x improvement (it has wrapped hash aggregator with a decimal column) with this fix. Release note (performance improvement): A performance inefficiency has been fixed in the vectorized execution engine which results in speed ups on all queries when run via the vectorized engine, with most noticeable gains on the queries that output many rows.
|
This change is |
asubiotto
approved these changes
May 13, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Backport 1/2 commits from #48669.
/cc @cockroachdb/release
colexec: fix performance inefficiency in materializer
We mistakenly were passing
sqlbase.DatumAllocby value, and not bypointer, and as a result we would always be allocating 16 datums but
using only 1 - i.e. we were not only not pooling the allocations, but
actually making a bunch of useless allocations as well.
This inefficiency becomes noticeable when the vectorized query returns
many rows and when we have wrapped processors and those processors get
a lot of input rows - in all cases when we need to materialize a lot.
For example, TPC-H query 16 sees about 10% improvement (it returns 18k
rows) and TPC-DS query 6 sees 2x improvement (it has wrapped hash
aggregator with a decimal column) with this fix.
Release note (performance improvement): A performance inefficiency has
been fixed in the vectorized execution engine which results in speed ups
on all queries when run via the vectorized engine, with most noticeable
gains on the queries that output many rows.