Nic Washbourne

compiler: Fix complex arguments and implement float16 lowering

3

Fixes some issues that prevented complex float arguments from being passed correctly. Also implements lowering for half-precision floats with the `_Float16` type. Some remaining tasks: - MPI with complex floating...

compiler: Cache `Dependence` instances

4

Adds a `@_memoized_instances` decorator for caching instances returned by `__new__`. This decorator is applied to `Relation` (and indirectly, `Dependence`) for a somewhat substantial speedup in build times. Tested locally (with...

compiler: Lazy IET visitors + Search

6

This PR implements lazy visitors in the IET layer which operate via generators instead of flattening lists of children's results at every node. Replaces `FindSymbols`, `FindNodes`, `FindWithin` and `FindApplications` with...

Rename `ir.clusters.visitors.Queue`

2

The `Queue` cluster visitor is confusingly named for something that performs a divide-and-conquer process; I'd propose renaming it to something but not sure what. Specifically making this issue because I'm...

docker: Add Python 3.13 free-threaded Docker images

8

Adds an additional base layer for building Python 3.13 from source with `disable-gil`, and makes the necessary changes to the rest of the build process as well as Docker build...

misc: Thread-safe cache rewrite

1

Rewrites `memoized_meth` and `memoized_generator` for concurrency, as well as dropping the legacy `memoized_func` in favor of `functools.cache`. Also adds a global lock for the symbol cache allowing thread-safe `Symbol` construction...