All pull requests

Pull requests

• cudax sharded: concepts for sharded structures + the concept-generic algorithm tier (pilot)

  #13·caugonnet opened Sep 2, 2026Owner·

  ·

• [sharded 7/7] CUDA graph capture: the contract, enforced and tested

  #11·caugonnet opened Aug 21, 2026Owner·

  ·

• cudax sharded: a shared-address-space engine for the places-rung sort

  #10·caugonnet opened Aug 21, 2026Owner·

  ·

• python cuda_sharded: bindings for the sharded rung (bound containers, tiered algorithms, free-threading-native)

  #9·caugonnet opened Aug 20, 2026Owner·

  ·

• [stf] Enable free-threaded Python (3.14t) support for the cuda.stf bindings

  #8·caugonnet opened Aug 20, 2026Owner·

  ·
  

• cudax sharded: sharded_csr + opt-in cuSPARSE sparse products (spmv/spmm) with measured rebalance

  #7·caugonnet opened Aug 20, 2026Owner·

  ·
  2

• [cudax] __sharded: sort -- the MGMN distributed sort as the engine behind the sharded name

  #6·caugonnet opened Aug 20, 2026Owner·

  ·

• [cudax] __sharded: places communicator (a places-backed model of the multi-GPU communicator concept) + the engine-bindings adapter

  #5·caugonnet opened Aug 20, 2026Owner·

  ·

• cudax: __sharded — placed containers and algorithms for the in-process rung

  #4·caugonnet opened Aug 20, 2026Owner·

  ·

• cudax places: place_group — a grid of places plus its execution resources

  #3·caugonnet opened Aug 20, 2026Owner·

  ·