Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

reduce allocations when binding string/time args #1293

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

charlievieth
Copy link
Contributor

@charlievieth charlievieth commented Nov 5, 2024

This commit reduces the number of allocations required to bind args by eliminating string to byte slice conversions for string and time.Time types and by only checking for bind parameters if any of the driver.NamedValue args are named. It also changes bind to only reset the sqlite3 statement when necessary - previously the statement was always reset even on first use (this yields a 3-4% performance boost).

goos: darwin
goarch: arm64
pkg: github.com/mattn/go-sqlite3
cpu: Apple M4 Pro
                         │   y1.txt    │               y2.txt               │
                         │   sec/op    │   sec/op     vs base               │
Suite/BenchmarkQuery-14    2.080µ ± 3%   1.995µ ± 0%  -4.13% (p=0.000 n=10)
Suite/BenchmarkParams-14   2.282µ ± 1%   2.181µ ± 2%  -4.43% (p=0.000 n=10)
Suite/BenchmarkStmt-14     1.537µ ± 1%   1.489µ ± 1%  -3.16% (p=0.000 n=10)
geomean                    1.939µ        1.864µ       -3.91%

                         │   y1.txt    │                y2.txt                 │
                         │    B/op     │    B/op      vs base                  │
Suite/BenchmarkQuery-14     688.0 ± 0%    688.0 ± 0%        ~ (p=1.000 n=10) ¹
Suite/BenchmarkParams-14   1104.0 ± 0%   1000.0 ± 0%   -9.42% (p=0.000 n=10)
Suite/BenchmarkStmt-14      920.0 ± 0%    816.0 ± 0%  -11.30% (p=0.000 n=10)
geomean                     887.4         824.9        -7.04%
¹ all samples are equal

                         │   y1.txt   │               y2.txt                │
                         │ allocs/op  │ allocs/op   vs base                 │
Suite/BenchmarkQuery-14    23.00 ± 0%   23.00 ± 0%       ~ (p=1.000 n=10) ¹
Suite/BenchmarkParams-14   27.00 ± 0%   25.00 ± 0%  -7.41% (p=0.000 n=10)
Suite/BenchmarkStmt-14     25.00 ± 0%   23.00 ± 0%  -8.00% (p=0.000 n=10)
geomean                    24.95        23.65       -5.20%
¹ all samples are equal

@charlievieth
Copy link
Contributor Author

charlievieth commented Nov 5, 2024

Note: this PR includes #1294 since it fixes the CIFuzz and macOS tests.

This is no longer true since that commit/pr are no longer needed to make the CI tests pass.

@charlievieth charlievieth force-pushed the cev/bind-perf branch 2 times, most recently from 011926c to 5959360 Compare November 28, 2024 22:57
@charlievieth charlievieth force-pushed the cev/bind-perf branch 4 times, most recently from 6db1eec to a7c0d28 Compare December 18, 2024 00:44
This commit reduces the number of allocations required to bind args by
eliminating string to byte slice conversions for string and time.Time
types and by only checking for bind parameters if any of the
driver.NamedValue args are named. It also changes bind to only reset the
sqlite3 statement when necessary - previously the statement was always
reset even on first use (this yields a 3-4% performance boost).

goos: darwin
goarch: arm64
pkg: github.com/mattn/go-sqlite3
cpu: Apple M4 Pro
                         │   y1.txt    │               y2.txt               │
                         │   sec/op    │   sec/op     vs base               │
Suite/BenchmarkQuery-14    2.080µ ± 3%   1.995µ ± 0%  -4.13% (p=0.000 n=10)
Suite/BenchmarkParams-14   2.282µ ± 1%   2.181µ ± 2%  -4.43% (p=0.000 n=10)
Suite/BenchmarkStmt-14     1.537µ ± 1%   1.489µ ± 1%  -3.16% (p=0.000 n=10)
geomean                    1.939µ        1.864µ       -3.91%

                         │   y1.txt    │                y2.txt                 │
                         │    B/op     │    B/op      vs base                  │
Suite/BenchmarkQuery-14     688.0 ± 0%    688.0 ± 0%        ~ (p=1.000 n=10) ¹
Suite/BenchmarkParams-14   1104.0 ± 0%   1000.0 ± 0%   -9.42% (p=0.000 n=10)
Suite/BenchmarkStmt-14      920.0 ± 0%    816.0 ± 0%  -11.30% (p=0.000 n=10)
geomean                     887.4         824.9        -7.04%
¹ all samples are equal

                         │   y1.txt   │               y2.txt                │
                         │ allocs/op  │ allocs/op   vs base                 │
Suite/BenchmarkQuery-14    23.00 ± 0%   23.00 ± 0%       ~ (p=1.000 n=10) ¹
Suite/BenchmarkParams-14   27.00 ± 0%   25.00 ± 0%  -7.41% (p=0.000 n=10)
Suite/BenchmarkStmt-14     25.00 ± 0%   23.00 ± 0%  -8.00% (p=0.000 n=10)
geomean                    24.95        23.65       -5.20%
¹ all samples are equal
@charlievieth
Copy link
Contributor Author

@mattn / @rittneje I know this time of year is pretty busy, but was wondering if either of you had the time to take a look at this PR?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant