perf: improve performance of string replace #19530
Merged
+268
−15
Conversation
This file contains hidden or 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
viirya
force-pushed
the
replace_optimize_v2
branch
2 times, most recently
from
3e29ea3 to
3f8c9bb
Compare
Use a reusable String buffer instead of allocating a new String for each row. This optimization achieves 17-32% performance improvement across different string types and sizes by avoiding per-row allocations. Benchmark results: | Benchmark | Array Size | String Length | Baseline (µs) | Optimized (µs) | Improvement | |---------------|------------|---------------|---------------|----------------|-------------| | string_view | 1024 | 32 | 32.53 | 22.32 | 31.4% faster| | string | 1024 | 32 | 31.89 | 21.49 | 32.6% faster| | large_string | 1024 | 32 | 31.75 | 22.01 | 30.7% faster| | string_view | 1024 | 128 | 49.51 | 36.11 | 27.1% faster| | string | 1024 | 128 | 48.91 | 34.90 | 28.6% faster| | large_string | 1024 | 128 | 49.78 | 35.42 | 28.8% faster| | string_view | 4096 | 32 | 133.67 | 95.93 | 28.2% faster| | string | 4096 | 32 | 131.48 | 91.73 | 30.2% faster| | large_string | 4096 | 32 | 129.61 | 92.82 | 28.4% faster| | string_view | 4096 | 128 | 191.50 | 153.74 | 19.7% faster| | string | 4096 | 128 | 185.27 | 149.37 | 19.4% faster| | large_string | 4096 | 128 | 187.82 | 154.32 | 17.8% faster|
viirya
force-pushed
the
replace_optimize_v2
branch
from
3f8c9bb to
c10d410
Compare
Jefffrey
reviewed
Dec 29, 2025
| Ok(Arc::new(result) as ArrayRef) | ||
| /// Helper function to perform string replacement into a reusable String buffer | ||
| #[inline] | ||
| fn replace_into_string(buffer: &mut String, string: &str, from: &str, to: &str) { |
Contributor
There was a problem hiding this comment.
I wonder if we lost other optimization tricks that the Rust replace function provides; for example looking at the code: https://doc.rust-lang.org/src/alloc/str.rs.html#268
It has a Fast path for replacing a single ASCII character with another. I wonder if we capture this in our benchmarks?
Member
Author
There was a problem hiding this comment.
Excellent point! Thanks for the review. The single ASCII character fast path optimization has been added.
viirya
force-pushed
the
replace_optimize_v2
branch
from
abdede8 to
39ae000
Compare
Add a fast path for replacing single ASCII characters with another single ASCII character, matching Rust's str::replace() optimization. This enables vectorization and avoids UTF-8 boundary checking overhead. Changes: - Added ASCII character detection in replace_into_string() - When both 'from' and 'to' are single ASCII bytes, use direct byte mapping - Updated benchmark to include single ASCII character replacement tests Optimization: - Fast path operates directly on bytes using simple map operation - Compiler can vectorize the byte-wise replacement - Avoids overhead of match_indices() pattern matching for this common case Benchmark Results (Single ASCII Character Replacement) against previous commit: - size=1024, str_len=32: 29.5 µs → 21.4 µs (27% faster) - size=1024, str_len=128: 73.9 µs → 23.4 µs (68% faster) - size=4096, str_len=32: 121.8 µs → 85.6 µs (30% faster) - size=4096, str_len=128: 316.9 µs → 83.8 µs (74% faster) The optimization shows exceptional 27-74% improvements, with the benefit scaling dramatically with string length. For 128-character strings, we achieve over 3x speedup by enabling vectorization and eliminating pattern matching overhead. This addresses reviewer feedback about capturing Rust's str::replace() optimization tricks for single ASCII character replacements.
viirya
force-pushed
the
replace_optimize_v2
branch
from
39ae000 to
4f36125
Compare
Member
Author
|
Thanks @andygrove @Jefffrey |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
Which issue does this PR close?
Rationale for this change
What changes are included in this PR?
Use a reusable String buffer instead of allocating a new String for each row. This optimization achieves 17-32% performance improvement across different string types and sizes by avoiding per-row allocations.
Benchmark results:
Are these changes tested?
Are there any user-facing changes?