The pdqsort has been merged into the Go standard library since Go 1.19, please use
sortorslicesdirectly instead of this package.issue: golang/go#50154
commit: https://github.com/golang/go/commit/72e77a7f41bbf45d466119444307fd3ae996e257
The algorithm is mainly based on pattern-defeating quicksort by Orson Peters.
Compared to sort.Ints(Go1.18), it is 2x faster in random slices, and 2x ~ 60x faster in common patterns.
- Paper: https://arxiv.org/pdf/2106.05123.pdf
- C++ implementation: https://github.com/orlp/pdqsort
- Rust implementation: https://docs.rs/pdqsort/latest/pdqsort/
Best Average Worst Memory Stable Deterministic
n n log n n log n log n No Yes
- Unstable sort, may reorder equal elements.
- Disable the optimization from BlockQuickSort, since its poor performance in Go.
package main
import (
"fmt"
"github.com/zhangyunhao116/pdqsort"
)
func main() {
x := []int{3, 1, 2, 4, 5, 9, 8, 7}
pdqsort.Slice(x)
fmt.Printf("%v\n", x)
}Go version: go1.18-a412b5f0d8 linux/amd64
CPU: Intel 11700k(8C16T)
OS: ubuntu 20.04
MEMORY: 16G x 2 (3200MHz)
name time/op
Random/pdqsort_64 1.18µs ± 0%
Random/stdsort_64 2.38µs ± 0%
Random/pdqsort_256 6.24µs ± 3%
Random/stdsort_256 13.2µs ± 7%
Random/pdqsort_1024 32.4µs ± 0%
Random/stdsort_1024 62.2µs ± 0%
Random/pdqsort_4096 149µs ± 0%
Random/stdsort_4096 291µs ± 0%
Random/pdqsort_65536 3.14ms ± 0%
Random/stdsort_65536 6.11ms ± 0%
Sorted/pdqsort_64 94.4ns ± 4%
Sorted/stdsort_64 711ns ± 0%
Sorted/pdqsort_256 171ns ± 1%
Sorted/stdsort_256 3.37µs ± 0%
Sorted/pdqsort_1024 507ns ± 0%
Sorted/stdsort_1024 16.6µs ± 0%
Sorted/pdqsort_4096 1.82µs ± 0%
Sorted/stdsort_4096 78.9µs ± 0%
Sorted/pdqsort_65536 28.2µs ± 0%
Sorted/stdsort_65536 1.73ms ± 0%
NearlySorted/pdqsort_64 353ns ± 2%
NearlySorted/stdsort_64 931ns ± 1%
NearlySorted/pdqsort_256 1.78µs ± 1%
NearlySorted/stdsort_256 4.99µs ± 1%
NearlySorted/pdqsort_1024 8.28µs ± 0%
NearlySorted/stdsort_1024 26.1µs ± 1%
NearlySorted/pdqsort_4096 38.2µs ± 0%
NearlySorted/stdsort_4096 117µs ± 1%
NearlySorted/pdqsort_65536 792µs ± 0%
NearlySorted/stdsort_65536 2.46ms ± 0%
Reversed/pdqsort_64 113ns ± 1%
Reversed/stdsort_64 845ns ± 0%
Reversed/pdqsort_256 253ns ± 1%
Reversed/stdsort_256 3.69µs ± 0%
Reversed/pdqsort_1024 785ns ± 0%
Reversed/stdsort_1024 17.8µs ± 1%
Reversed/pdqsort_4096 2.89µs ± 0%
Reversed/stdsort_4096 83.7µs ± 0%
Reversed/pdqsort_65536 45.1µs ± 0%
Reversed/stdsort_65536 1.80ms ± 0%
NearlyReversed/pdqsort_64 435ns ± 2%
NearlyReversed/stdsort_64 1.33µs ± 1%
NearlyReversed/pdqsort_256 2.13µs ± 1%
NearlyReversed/stdsort_256 7.23µs ± 1%
NearlyReversed/pdqsort_1024 10.6µs ± 1%
NearlyReversed/stdsort_1024 38.4µs ± 1%
NearlyReversed/pdqsort_4096 50.2µs ± 1%
NearlyReversed/stdsort_4096 176µs ± 0%
NearlyReversed/pdqsort_65536 1.04ms ± 1%
NearlyReversed/stdsort_65536 3.57ms ± 0%
Mod8/pdqsort_64 345ns ± 2%
Mod8/stdsort_64 949ns ± 1%
Mod8/pdqsort_256 1.02µs ± 1%
Mod8/stdsort_256 3.78µs ± 0%
Mod8/pdqsort_1024 3.13µs ± 2%
Mod8/stdsort_1024 14.3µs ± 0%
Mod8/pdqsort_4096 10.2µs ± 1%
Mod8/stdsort_4096 59.3µs ± 0%
Mod8/pdqsort_65536 190µs ± 2%
Mod8/stdsort_65536 1.00ms ± 0%
AllEqual/pdqsort_64 89.4ns ± 3%
AllEqual/stdsort_64 381ns ± 1%
AllEqual/pdqsort_256 170ns ± 1%
AllEqual/stdsort_256 1.03µs ± 1%
AllEqual/pdqsort_1024 507ns ± 0%
AllEqual/stdsort_1024 3.66µs ± 0%
AllEqual/pdqsort_4096 1.82µs ± 0%
AllEqual/stdsort_4096 14.0µs ± 0%
AllEqual/pdqsort_65536 28.2µs ± 0%
AllEqual/stdsort_65536 224µs ± 0%