1 dépôt
Enforces restrictions on parallel loops over local fragment buffers, such as requiring non-symbolic loop ranges.
Distinct from Parallel Loop Transformers: Distinct from Parallel Loop Transformers: focuses on validation of loop constraints over fragment buffers, not general loop transformation.
Explore 1 awesome GitHub repository matching data & databases · Fragment Buffer Loop Validators. Refine with filters or upvote what's useful.
TileLang is a Python-embedded domain-specific language compiler that JIT-compiles and autotunes GPU kernels. It uses a tile-based DSL, automatic software pipelining, and parallel autotuning to generate optimized GPU kernels at runtime. It supports tensor core operations with Pythonic syntax, automatic memory management, and thread mapping. The compiler searches over tile sizes, thread counts, and scheduling policies, compiling and benchmarking candidates in parallel to find the fastest kernel. It also caches compiled binaries and tuning results to disk for reuse across sessions. TileLang inc
Enforces restrictions on parallel loops over local fragment buffers to ensure correctness in GPU kernels.