3 ms·
Faster, but still 15-25% slower than Highway's vqsort according to my somewhat amateur tests: ------------------------------------------------------------
by sagarm 3y ago
Faster, but still 15-25% slower than Highway's vqsort according to my somewhat amateur tests:
------------------------------------------------------------
Benchmark Time Bytes Items
------------------------------------------------------------
Hwy/random 2.16 s 474.1M/s 62.1M/s
Hwy/sorted_runs 2.29 s 446.7M/s 58.5M/s
IntelX86SIMD/random 2.90 s 353.9M/s 46.3M/s
IntelX86SIMD/sorted_runs 2.68 s 382.4M/s 50.1M/s
https://github.com/funrollloops/parallel-sort-bench https://github.com/funrollloops/parallel-sort-bench
- eesmith 3y agoSee also "10~17x faster than what? A performance analysis of Intel's x86-simd-sort (AVX-512)" at https://github.com/Voultapher/sort-research-rs/blob/main/writeup/intel_avx512/text.md https://github.com/Voultapher/sort-research-rs/blob/main/wri... posted here at https://news.ycombinator.com/item?id=36273544 https://news.ycombinator.com/item?id=36273544 and minor additional comments at https://news.ycombinator.com/item?id=36268877 https://news.ycombinator.com/item?id=36268877 . When those benchmarks were first done, for random input, vqsort was faster once you get around ~20,000 items, but for 1,000 items the new AVX512 sort is 13x faster than vqsort. As you read it, you'll see that the vqsort issue for small arrays has been fixed, and as of a few weeks ago or so, vqsort is now faster than the AVX512 sort for random.