reduce CGO calls when scanning rows #1291
Open
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
charlievieth
changed the title
charlievieth
changed the title
charlievieth
changed the title
charlievieth
changed the title
|
Commit aa590cf addresses an issue with |
This commit improves the performance of queries by at least 20% by only
collecting the information needed to convert sqlite3 values to Go values
and by batching CGO calls when scanning rows (the performance
improvement scales with the number of columns being scanned).
This commit adds a new coltype field to the SQLiteRows struct which
stores the declared column type (either data/time or boolean) and the
sqlite3 datatype. Previously, this library would fetch the string
representation of each column, which is inefficient and rarely needed
since the non-standard SQLiteRows.DeclTypes method is rarely called.
It also changes the benchmark suite to use an in-memory database since
we do not want the file system interfering with benchmark results.
goos: darwin
goarch: arm64
pkg: github.com/mattn/go-sqlite3
cpu: Apple M1 Max
│ base.10.txt │ new.10.txt │
│ sec/op │ sec/op vs base │
CustomFunctions-10 3.318µ ± 2% 3.115µ ± 2% -6.10% (p=0.000 n=10)
Suite/BenchmarkExec-10 1.236µ ± 1% 1.240µ ± 2% ~ (p=0.617 n=10)
Suite/BenchmarkQuery-10 4.004µ ± 7% 3.363µ ± 2% -16.02% (p=0.000 n=10)
Suite/BenchmarkParams-10 4.241µ ± 1% 3.758µ ± 2% -11.40% (p=0.000 n=10)
Suite/BenchmarkStmt-10 2.830µ ± 0% 2.378µ ± 2% -15.97% (p=0.000 n=10)
Suite/BenchmarkRows-10 126.3µ ± 1% 101.3µ ± 1% -19.79% (p=0.000 n=10)
Suite/BenchmarkStmtRows-10 124.9µ ± 1% 100.5µ ± 2% -19.56% (p=0.000 n=10)
Suite/BenchmarkStmt10Cols-10 10.130µ ± 0% 7.042µ ± 1% -30.48% (p=0.000 n=10)
geomean 8.655µ 7.328µ -15.33%
│ base.10.txt │ new.10.txt │
│ B/op │ B/op vs base │
CustomFunctions-10 568.0 ± 0% 576.0 ± 0% +1.41% (p=0.000 n=10)
Suite/BenchmarkExec-10 128.0 ± 0% 128.0 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkQuery-10 688.0 ± 0% 648.0 ± 0% -5.81% (p=0.000 n=10)
Suite/BenchmarkParams-10 1.078Ki ± 0% 1.031Ki ± 0% -4.35% (p=0.000 n=10)
Suite/BenchmarkStmt-10 920.0 ± 0% 872.0 ± 0% -5.22% (p=0.000 n=10)
Suite/BenchmarkRows-10 9.305Ki ± 0% 9.188Ki ± 0% -1.26% (p=0.000 n=10)
Suite/BenchmarkStmtRows-10 9.289Ki ± 0% 9.164Ki ± 0% -1.35% (p=0.000 n=10)
Suite/BenchmarkStmt10Cols-10 992.0 ± 0% 696.0 ± 0% -29.84% (p=0.000 n=10)
geomean 1.181Ki 1.106Ki -6.35%
¹ all samples are equal
│ base.10.txt │ new.10.txt │
│ allocs/op │ allocs/op vs base │
CustomFunctions-10 18.00 ± 0% 18.00 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkExec-10 7.000 ± 0% 7.000 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkQuery-10 23.00 ± 0% 23.00 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkParams-10 27.00 ± 0% 27.00 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkStmt-10 25.00 ± 0% 25.00 ± 0% ~ (p=1.000 n=10) ¹
Suite/BenchmarkRows-10 525.0 ± 0% 519.0 ± 0% -1.14% (p=0.000 n=10)
Suite/BenchmarkStmtRows-10 524.0 ± 0% 518.0 ± 0% -1.15% (p=0.000 n=10)
Suite/BenchmarkStmt10Cols-10 39.00 ± 0% 19.00 ± 0% -51.28% (p=0.000 n=10)
geomean 46.26 42.17 -8.86%
¹ all samples are equal
charlievieth
force-pushed
the
cev/col-perf
branch
from
aa590cf
to
fd9d096
Compare
|
Codecov ReportAttention: Patch coverage is
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## master #1291 +/- ##
==========================================
- Coverage 47.16% 47.14% -0.02%
==========================================
Files 12 12
Lines 1533 1542 +9
==========================================
+ Hits 723 727 +4
- Misses 669 672 +3
- Partials 141 143 +2 ☔ View full report in Codecov by Sentry. |
charlievieth
force-pushed
the
cev/col-perf
branch
from
fd9d096
to
c65c946
Compare
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.
This commit improves the performance of queries by at least 15% by only collecting the information needed to convert sqlite3 values to Go values and by batching CGO calls when scanning rows (the performance improvement scales with the number of columns being scanned).
This commit adds a new coltype field to the SQLiteRows struct which stores the declared column type (either data/time or boolean) and the sqlite3 datatype. Previously, this library would fetch the string representation of each column, which is inefficient and rarely needed since the non-standard SQLiteRows.DeclTypes method is rarely called.
It also changes the benchmark suite to use an in-memory database since we do not want the file system interfering with benchmark results.