Vortex quantization and topk kernel adaption#1

Open

zxr-creator wants to merge 17 commits intoInfini-AI-Lab:v1from

Commits on Feb 22, 2026

Add int8 quantization for vortex.
zxr-creator
committed

Commits on Feb 23, 2026

1. Add support for pro 6000. 2. Correction for vortex
zxr-creator
committed
1. Correction on int8 (maximize memory occupation) 2. Implement fp8 quantization.
zxr-creator
committed

Commits on Mar 1, 2026

update on parameters for reduce_pp_kernel with quantization
zxr-creator
committed

Commits on Mar 2, 2026

adapt topk kernel from sglang to vortex
zxr-creator
committed

Commits on Mar 9, 2026

Commits on Mar 18, 2026

add aim24
zxr-creator
committed

Commits on Mar 24, 2026

Commits on Mar 29, 2026

fix on the ragged warpper, using single ragged warpper on concated rags and pages; fix on the previous quantization implementaion, with lanuch_graph dtype set to the quant type
zxr-creator
committed

Commits on Mar 30, 2026

Commits on Mar 31, 2026

Add TopK benchmarking suite and related scripts
zxr-creator
committed

Commits on Apr 1, 2026

Enhance TopK mapping modes with new remap functions
zxr-creator
committed

Commits on Apr 2, 2026

enhance TopK mapping with adaptive tail-window mode; modify example scripts to reflect changes in histogram calibration and TopK mapping parameters.
zxr-creator
committed

Commits on Apr 7, 2026

Enhance TopK mapping with new modes and original sglang kernel support
zxr-creator
committed