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

perf: avoid allocs in unicodeFoldTransformer #53

Merged
merged 1 commit into from
May 3, 2023

Conversation

josharian
Copy link
Contributor

There is no need for temporary slices; the transformation can happen rune-by-rune.

goos: darwin
goarch: arm64
pkg: github.com/lithammer/fuzzysearch/fuzzy
                              │      a       │                  b                  │
                              │    sec/op    │   sec/op     vs base                │
Match-8                          16.06n ± 3%   16.04n ± 3%        ~ (p=0.447 n=10)
MatchBigLate-8                   1.008µ ± 0%   1.007µ ± 1%        ~ (p=1.000 n=10)
MatchBigEarly-8                  12.56n ± 2%   12.58n ± 3%        ~ (p=0.303 n=10)
MatchFold-8                      334.5n ± 1%   137.9n ± 2%  -58.77% (p=0.000 n=10)
MatchFoldBigLate-8              27.349µ ± 1%   7.088µ ± 1%  -74.08% (p=0.000 n=10)
MatchFoldBigEarly-8             26.390µ ± 2%   6.091µ ± 9%  -76.92% (p=0.000 n=10)
RankMatch-8                      17.66n ± 1%   17.59n ± 0%   -0.40% (p=0.034 n=10)
RankMatchBigLate-8               1.010µ ± 1%   1.008µ ± 3%        ~ (p=0.368 n=10)
RankMatchBigEarly-8              1.236µ ± 0%   1.235µ ± 2%        ~ (p=0.667 n=10)
LevenshteinDistance-8            55.65n ± 3%   55.52n ± 3%        ~ (p=0.054 n=10)
LevenshteinDistanceBigLate-8     20.53µ ± 1%   20.47µ ± 3%   -0.29% (p=0.042 n=10)
LevenshteinDistanceBigEarly-8    20.57µ ± 3%   20.47µ ± 3%        ~ (p=0.060 n=10)
geomean                          736.4n        540.1n       -26.66%

                              │        a        │                   b                    │
                              │      B/op       │     B/op      vs base                  │
Match-8                            0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigLate-8                     0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigEarly-8                    0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchFold-8                        720.0 ± 0%       512.0 ± 0%  -28.89% (p=0.000 n=10)
MatchFoldBigLate-8              36.711Ki ± 0%     8.750Ki ± 0%  -76.17% (p=0.000 n=10)
MatchFoldBigEarly-8             36.711Ki ± 0%     8.750Ki ± 0%  -76.17% (p=0.000 n=10)
RankMatch-8                        0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigLate-8                 0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigEarly-8                0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistance-8              0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigLate-8     6.375Ki ± 0%     6.375Ki ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigEarly-8    6.375Ki ± 0%     6.375Ki ± 0%        ~ (p=1.000 n=10) ¹
geomean                                       ²                 -23.46%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                              │       a       │                  b                   │
                              │   allocs/op   │ allocs/op   vs base                  │
Match-8                          0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigLate-8                   0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigEarly-8                  0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchFold-8                     10.000 ± 0%     2.000 ± 0%  -80.00% (p=0.000 n=10)
MatchFoldBigLate-8              186.00 ± 0%     25.00 ± 0%  -86.56% (p=0.000 n=10)
MatchFoldBigEarly-8             186.00 ± 0%     25.00 ± 0%  -86.56% (p=0.000 n=10)
RankMatch-8                      0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigLate-8               0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigEarly-8              0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistance-8            0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigLate-8     1.000 ± 0%     1.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigEarly-8    1.000 ± 0%     1.000 ± 0%        ~ (p=1.000 n=10) ¹
geomean                                     ²               -37.41%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

There is no need for temporary slices;
the transformation can happen rune-by-rune.

goos: darwin
goarch: arm64
pkg: github.com/lithammer/fuzzysearch/fuzzy
                              │      a       │                  b                  │
                              │    sec/op    │   sec/op     vs base                │
Match-8                          16.06n ± 3%   16.04n ± 3%        ~ (p=0.447 n=10)
MatchBigLate-8                   1.008µ ± 0%   1.007µ ± 1%        ~ (p=1.000 n=10)
MatchBigEarly-8                  12.56n ± 2%   12.58n ± 3%        ~ (p=0.303 n=10)
MatchFold-8                      334.5n ± 1%   137.9n ± 2%  -58.77% (p=0.000 n=10)
MatchFoldBigLate-8              27.349µ ± 1%   7.088µ ± 1%  -74.08% (p=0.000 n=10)
MatchFoldBigEarly-8             26.390µ ± 2%   6.091µ ± 9%  -76.92% (p=0.000 n=10)
RankMatch-8                      17.66n ± 1%   17.59n ± 0%   -0.40% (p=0.034 n=10)
RankMatchBigLate-8               1.010µ ± 1%   1.008µ ± 3%        ~ (p=0.368 n=10)
RankMatchBigEarly-8              1.236µ ± 0%   1.235µ ± 2%        ~ (p=0.667 n=10)
LevenshteinDistance-8            55.65n ± 3%   55.52n ± 3%        ~ (p=0.054 n=10)
LevenshteinDistanceBigLate-8     20.53µ ± 1%   20.47µ ± 3%   -0.29% (p=0.042 n=10)
LevenshteinDistanceBigEarly-8    20.57µ ± 3%   20.47µ ± 3%        ~ (p=0.060 n=10)
geomean                          736.4n        540.1n       -26.66%

                              │        a        │                   b                    │
                              │      B/op       │     B/op      vs base                  │
Match-8                            0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigLate-8                     0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigEarly-8                    0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchFold-8                        720.0 ± 0%       512.0 ± 0%  -28.89% (p=0.000 n=10)
MatchFoldBigLate-8              36.711Ki ± 0%     8.750Ki ± 0%  -76.17% (p=0.000 n=10)
MatchFoldBigEarly-8             36.711Ki ± 0%     8.750Ki ± 0%  -76.17% (p=0.000 n=10)
RankMatch-8                        0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigLate-8                 0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigEarly-8                0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistance-8              0.000 ± 0%       0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigLate-8     6.375Ki ± 0%     6.375Ki ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigEarly-8    6.375Ki ± 0%     6.375Ki ± 0%        ~ (p=1.000 n=10) ¹
geomean                                       ²                 -23.46%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                              │       a       │                  b                   │
                              │   allocs/op   │ allocs/op   vs base                  │
Match-8                          0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigLate-8                   0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchBigEarly-8                  0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
MatchFold-8                     10.000 ± 0%     2.000 ± 0%  -80.00% (p=0.000 n=10)
MatchFoldBigLate-8              186.00 ± 0%     25.00 ± 0%  -86.56% (p=0.000 n=10)
MatchFoldBigEarly-8             186.00 ± 0%     25.00 ± 0%  -86.56% (p=0.000 n=10)
RankMatch-8                      0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigLate-8               0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
RankMatchBigEarly-8              0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistance-8            0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigLate-8     1.000 ± 0%     1.000 ± 0%        ~ (p=1.000 n=10) ¹
LevenshteinDistanceBigEarly-8    1.000 ± 0%     1.000 ± 0%        ~ (p=1.000 n=10) ¹
geomean                                     ²               -37.41%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean
@lithammer lithammer merged commit 6561906 into lithammer:master May 3, 2023
@lithammer
Copy link
Owner

Very nice, thank you!

@lithammer
Copy link
Owner

Released as part of v1.1.7

@josharian josharian deleted the optimize branch May 15, 2023 16:38
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.

2 participants