-
Notifications
You must be signed in to change notification settings - Fork 13.6k
Closed
Labels
I-slowIssue: Problems and improvements with respect to performance of generated code.Issue: Problems and improvements with respect to performance of generated code.
Description
Annoying issues, but not hard blockers:
- split Index into Index, IndexMut and IndexAssign
- make Vec implement the slice methods
- Constructing an iterator from a slice doesn't optimise completely
This is a significant performance improvement for any vectors that are grown, and will reduce heap allocation since zero-length vectors do not allocate and it does not have a header.
test bench::old_push ... bench: 43200 ns/iter (+/- 804)
test bench::old_push_preallocated ... bench: 42948 ns/iter (+/- 4043)
test bench::push ... bench: 7190 ns/iter (+/- 245)
test bench::push_preallocated ... bench: 6606 ns/iter (+/- 170)
#[cfg(test)]
extern mod extra;
#[cfg(test)]
mod bench {
use std;
use std::vec_ng::Vec;
use extra::test::BenchHarness;
static size: uint = 10000;
#[bench]
fn push(bh: &mut BenchHarness) {
bh.iter(|| {
let mut xs = Vec::with_capacity(16);
for i in range(0, size) {
xs.push(i);
}
});
}
#[bench]
fn old_push(bh: &mut BenchHarness) {
bh.iter(|| {
let mut xs = std::vec::with_capacity(16);
for i in range(0, size) {
xs.push(i);
}
});
}
#[bench]
fn push_preallocated(bh: &mut BenchHarness) {
bh.iter(|| {
let mut xs = Vec::with_capacity(size);
for i in range(0, size) {
xs.push(i);
}
});
}
#[bench]
fn old_push_preallocated(bh: &mut BenchHarness) {
bh.iter(|| {
let mut xs = std::vec::with_capacity(size);
for i in range(0, size) {
xs.push(i);
}
});
}
}
I moved the priority tag here from the old issue, as this is the same core issue but without the distractions of past issues and discussions that are no longer relevant.
Metadata
Metadata
Assignees
Labels
I-slowIssue: Problems and improvements with respect to performance of generated code.Issue: Problems and improvements with respect to performance of generated code.