Skip to content

Add 16WPMSP: parallel AVX-512 int32 sort - #1909

Open
TylerBurger777 wants to merge 1 commit into
fffaraz:masterfrom
TylerBurger777:add-16wpmsp
Open

TylerBurger777 wants to merge 1 commit into
fffaraz:masterfrom
TylerBurger777:add-16wpmsp

Conversation

@TylerBurger777

Copy link
Copy Markdown

Adding 16WPMSP, a parallel int32 sort built on AVX-512 + std::thread.
Radix + 16-way merge tree with adaptive thread count and auto core detection.

Benchmarks (Ryzen 9 7950X, isolated binaries, median of 21):

  • 1M: 1.82 ms (2.5x faster than IPS4o)
  • 4M: 4.49 ms (1.6x faster than IPS4o, 2.0x faster than TBB)
  • 16M: 20.4 ms

Full results and honest crossover analysis:
https://github.com/TylerBurger777/16WPMSP/blob/main/docs/RESULTS.md

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants