Skip to content

Optimize ISE decoding - #785

Merged
solidpixel merged 1 commit into
mainfrom
opt-ise-quant-specialization
Nov 17, 2026
Merged

solidpixel merged 1 commit into
mainfrom
opt-ise-quant-specialization

Conversation

@solidpixel

@solidpixel solidpixel commented Nov 17, 2026 •

Copy link
Copy Markdown
Contributor

LZ RDO stresses the ISE decoding as it has to decode bit patterns a lot. Normal compression doesn't do this at all, so we've not optimized it much before.

This patch uses compile-time constexpr specialization and templates to generate specialized decode functions for each ISE quant level. This makes it feasible to map relatively cleanly to bit extract/compress/deposit ISA extensions for both x86-64 BMI/BMI2/VBMI and Arm SVE2+BitPerm.

RDO performance improves by around 2%, normal decompressor performance increases from ~640MT/s to 750MT/s.

@solidpixel
solidpixel merged commit 2349416 into main Nov 17, 2026
22 checks passed
@solidpixel
solidpixel deleted the opt-ise-quant-specialization branch November 17, 2026 20:18
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.

1 participant