Commits

Commits on Jun 16, 2023

re-enable tree freeing and context zeroing
wolfwood
committedJun 16, 2023
proofreading
wolfwood
committedJun 16, 2023
remove extraneous code
wolfwood
committedJun 16, 2023
rename First Pass for clarity
wolfwood
committedJun 16, 2023
comment edits
wolfwood
committedJun 16, 2023
rename for clarity
wolfwood
committedJun 16, 2023

Commits on Dec 8, 2022

provide background for the problem domain
wolfwood
committedDec 8, 2022
fix for gdc compilation
wolfwood
committedDec 8, 2022
file reorg
wolfwood
committedDec 8, 2022

Commits on Nov 20, 2022

Merge remote-tracking branch 'origin/rust' into feature/cleanup
wolfwood
committedNov 20, 2022
update for zig compiler 0.10.0
wolfwood
committedNov 20, 2022

Commits on Nov 14, 2022

Nodes and Layouts are now structs so they can have traits
wolfwood
committedNov 14, 2022

Commits on Aug 4, 2022

c-style node/layout conversions. rewrite more idiomatically
wolfwood
committedAug 4, 2022

Commits on Feb 7, 2022

use inline asm for smallestInLayout in zig
wolfwood
committedFeb 7, 2022

Commits on Jan 20, 2022

learned about inline avx512 asm in zig
wolfwood
committedJan 20, 2022
move rust to a project
wolfwood
committedJan 20, 2022

Commits on Jan 18, 2022

.gitignore
wolfwood
committedJan 18, 2022
zig implemention of a non-combinadic algo; beats combinadic C
wolfwood
committedJan 18, 2022

Commits on Jan 5, 2022

working zig implementation
wolfwood
committedJan 5, 2022

Commits on Dec 12, 2021

more c++ attempts
wolfwood
committedDec 12, 2021

Commits on Oct 17, 2021

emergency backup
wolfwood
committedOct 17, 2021

Commits on Apr 12, 2020

more C iteration versions, comments, and macros for selecting them
wolfwood
committedApr 12, 2020
beginnings of rust implemetation
wolfwood
committedApr 12, 2020
updated gitignore
wolfwood
committedApr 12, 2020

Commits on Apr 11, 2020

old stuff plus TODO
wolfwood
committedApr 11, 2020

Commits on Aug 12, 2018

just doing the work in a different place shaves off 8%... compiler must be doing some sweet loop-invariant code motion
wolfwood
committedAug 12, 2018

Commits on Aug 8, 2018

use mmap for giant Layout arrays. a bit faster, maybe because the GC doesn't copy them around?
wolfwood
committedAug 8, 2018
l2assert
wolfwood
committedAug 8, 2018
the combinadic deltas are faster than before/after arrays but this will be easier to refactor into the walk iterator
wolfwood
committedAug 8, 2018

Commits on Aug 7, 2018

D lang 10% speedup from using static square array for coeffs, instead of dynamic triangular one
wolfwood
committedAug 7, 2018

Commits on Aug 4, 2018

D lang rewrite. still needs mmap buffers and trimmed score lengths for intermediate zone
wolfwood
committedAug 4, 2018

Commits on Jun 30, 2017

faster direct lookup by amortizing costs across all decendents instead of calculating each combinadic rank individually
wolfwood
committedJun 30, 2017

Commits on Jun 28, 2017

cleanup
wolfwood
committedJun 28, 2017
only normalize once per unique score, instead of once per layout, moving normalization from the leaf functions to the main loop bodies
wolfwood
committedJun 28, 2017
now that memory usage has shrunk dramatically, default to anonymous mmap's instead of file-backed
wolfwood
committedJun 28, 2017