perf: store/cachekv: preallocate kvL in dirtyItems which gets appended too #14168
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
yihuang
approved these changes
Dec 6, 2022
yihuang
reviewed
Dec 6, 2022
…d too
Noticed in RAM profiles that the append for kvL had a concentrated
amount of RAM consumption yet where we invoke make didn't have any
capacity. Alas on profiling the results show a distribution which
eases on allocation pressure and distributes allocations across
different sections:
* Before:
```shell
. . 342: kvL := make([]*kv.Pair, 0)
. . 343: for i := startIndex; i <= endIndex; i++ {
. . 344: key := strL[i]
. . 345: cacheValue := store.cache[key]
1.14GB 1.14GB 346: kvL = append(kvL, &kv.Pair{Key: []byte(key), Value: cacheValue.value})
. . 347: }
```
* After:
```shell
76.33MB 76.33MB 342: kvL := make([]*kv.Pair, 0, endIndex-startIndex)
. . 343: for i := startIndex; i <= endIndex; i++ {
. . 344: key := strL[i]
. . 345: cacheValue := store.cache[key]
851.42MB 851.42MB 346: kvL = append(kvL, &kv.Pair{Key: []byte(key), Value: cacheValue.value})
. . 347: }
```
which also shows up in benchmark results:
```shell
$ benchstat before.txt after.txt
name old time/op new time/op delta
IteratorOnParentWith1MDeletes-8 1.69s ± 5% 1.65s ± 6% ~ (p=0.095 n=9+10)
name old alloc/op new alloc/op delta
IteratorOnParentWith1MDeletes-8 238MB ± 0% 214MB ± 0% -9.94% (p=0.000 n=10+9)
name old allocs/op new allocs/op delta
IteratorOnParentWith1MDeletes-8 3.10M ± 0% 3.10M ± 0% -0.00% (p=0.000 n=10+10)
```
Fixes #14167
odeke-em
force-pushed
the
store-cachekv-store-preallocate-kvL-in-dirtyItems
branch
from
32dbd02 to
1a14565
Compare
|
[Cosmos SDK] Kudos, SonarCloud Quality Gate passed! |
tac0turtle
added
the
A:automerge
tac0turtle
deleted the
store-cachekv-store-preallocate-kvL-in-dirtyItems
branch
BrandonWeng
added a commit
to sei-protocol/sei-cosmos
that referenced
this pull request
Dec 15, 2022
## Describe your changes and provide context Cherry picking changes here: * [\#14168](cosmos/cosmos-sdk#14168) perf: store/cachekv: preallocate kvL in dirtyItems which gets appended too * [\#10024](cosmos/cosmos-sdk#10024) fix!: store/cachekv: reduce growth factor for iterator ranging using binary searches #10024 ## Testing performed to validate your change Ran a cluster 
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.
Noticed in RAM profiles that the append for kvL had a concentrated amount of RAM consumption yet where we invoke make didn't have any capacity. Alas on profiling the results show a distribution which eases on allocation pressure and distributes allocations across different sections:
which also shows up in benchmark results:
$ benchstat before.txt after.txt name old time/op new time/op delta IteratorOnParentWith1MDeletes-8 1.69s ± 5% 1.65s ± 6% ~ (p=0.095 n=9+10) name old alloc/op new alloc/op delta IteratorOnParentWith1MDeletes-8 238MB ± 0% 214MB ± 0% -9.94% (p=0.000 n=10+9) name old allocs/op new allocs/op delta IteratorOnParentWith1MDeletes-8 3.10M ± 0% 3.10M ± 0% -0.00% (p=0.000 n=10+10)Fixes #14167