GPU offload for Nassau's Milnor multiply (p = 2) - #264

Closed

JoeyBF wants to merge 10 commits into

SpectralSequences:masterfrom

JoeyBF:nassau_gpu

Commits on Jul 11, 2026

Add a criterion benchmark suite for the algebra crate

JoeyBF
and
claude
committed
Add an admissible-matrix Milnor multiply at p = 2

JoeyBF
and
claude
committed
Add a Nassau end-to-end timing example and a Milnor multiply profiler

JoeyBF
and
claude
committed
Add hash-free seqno index tables for the Milnor multiply

JoeyBF
and
claude
committed
Add CubeCL GPU offload for Nassau's Milnor multiply

JoeyBF
and
claude
committed
Optimize the GPU Milnor multiply: reuse, caching, and memory

JoeyBF
and
claude
committed
Fix lint and rustdoc warnings in the GPU code

JoeyBF
and
claude
committed
Remove the compile-time Milnor multiply profiler

JoeyBF
and
claude
committed
Trim development-history comments from the GPU code

JoeyBF
and
claude
committed
Address CodeRabbit review: seqno-table race, u16 checks, invariants

JoeyBF
and
claude
committed