TECH RADAR / KERNELS & COMPILERS
CUTLASS
CUDA building blocks for matrix computation.
What it covers
GEMM, mixed precision, data movement, and CuTe abstractions.
What we would check
Document shapes, layouts, data types, architecture, and fusion boundaries before drawing performance conclusions.
Go to the source
A curated ecosystem entry. The evaluation guidance is our interpretation, not a report of a completed test or a performance ranking.