perf: fix O(n²) string allocations in fold operations #2257
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
Collaborator
|
Hi, @rnkrtt! Do you have a benchmark that shows any improvement with this change? |
Author
|
@gabrielbosio hey! Consistent ~3.5x speedup across different data sizes. Benchmark code (click)use std::time::Instant;
use std::fmt::Write;
fn old_way(data: &[u8]) -> String {
data.iter()
.fold(String::new(), |acc, b| acc + &format!("{:02x}", b))
}
fn new_way(data: &[u8]) -> String {
data.iter().fold(
String::with_capacity(data.len() * 2),
|mut string, b| {
write!(&mut string, "{:02x}", b).unwrap();
string
},
)
}
fn bench(name: &str, size: usize, iterations: usize, f: fn(&[u8]) -> String) {
let data: Vec<u8> = (0..size).map(|i| (i % 256) as u8).collect();
// Warmup
for _ in 0..5 {
let _ = f(&data);
}
// Actual measurement
let start = Instant::now();
for _ in 0..iterations {
let _ = f(&data);
}
let total_duration = start.elapsed();
let avg_duration = total_duration / iterations as u32;
println!("{} (size={}): {:?} avg over {} runs", name, size, avg_duration, iterations);
}
fn main() {
println!("Testing string fold performance...\n");
let configs = [
(100, 10000), // size, iterations
(1000, 1000),
(5000, 500),
(10000, 200),
];
for (size, iterations) in configs {
println!("--- Size: {} bytes ---", size);
bench("Old (fold + concat)", size, iterations, old_way);
bench("New (fold + write!) ", size, iterations, new_way);
// Calculate speedup
let data: Vec<u8> = (0..size).map(|i| (i % 256) as u8).collect();
let start = Instant::now();
for _ in 0..iterations { let _ = old_way(&data); }
let old_time = start.elapsed();
let start = Instant::now();
for _ in 0..iterations { let _ = new_way(&data); }
let new_time = start.elapsed();
let speedup = old_time.as_secs_f64() / new_time.as_secs_f64();
println!("Speedup: {:.2}x faster\n", speedup);
}
}Run with: |
Collaborator
|
Collaborator
|
Benchmark Results for unmodified programs 🚀
|
Collaborator
|
Benchmarks of the repo don't show improvements so it would be great if you can add a benchmark to see how much this change improves execution time in context |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Description
Fixes O(n²) string allocations in fold operations
Found two places where we're doing string concatenation inside fold, which creates a new string allocation on every iteration. This gets really slow with large data.
Changes:
for hex serialization
reference lists
Both were doing the same anti-pattern just in different places