All pull requests

Pull requests

• Bump the python-dependencies group with 10 updates

  dependenciespython:uv
  #2772·dependabot[bot] opened Nov 1, 2026Bot·

  ·

• Einsum: rebuild a contraction step with all of its operands

  #2771·abhinav-phi opened Oct 31, 2026·

  ·

• Lower the consume scope to a map before code generation

  #2770·tbennun opened Oct 31, 2026Collaborator·

  ·
  1

• [WIP] Readable experimental CPU code generator as the default

  #2769·ThrudPrimrose opened Oct 30, 2026Collaborator·

  ·

• Add random function

  #2768·twicki opened Oct 29, 2026·

  ·
  1

• SubgraphFusion: handle empty memlets and shared dynamic map range inputs

  #2766·ThrudPrimrose opened Oct 28, 2026Collaborator·

  ·

• Lower fractional and integer powers correctly in symstr and interstate edges

  #2764·ThrudPrimrose opened Oct 28, 2026Collaborator·

  ·

• Delste/synchronous tma

  #2751·DJAntivenom opened Oct 27, 2026·

  ·

• Cast-on-copy for CopyLibraryNode and CastNestedSDFGArguments pass

  #2746·ThrudPrimrose opened Oct 26, 2026Collaborator·

  ·
  1

  1

• [codegen] Generate multiple translation units with control flow regions

  #2742·tbennun opened Oct 25, 2026Collaborator·

  ·
  3

• [New GPU Codegen Prep] Add LowerDynamicThreadBlockMaps

  #2732·ThrudPrimrose opened Oct 22, 2026Collaborator·

  ·

• Round halfway cases to even in numpy.rint

  #2729·Shakhtar-Sankur opened Oct 22, 2026·

  ·
  2

• Lean

  #2718·tbennun opened Oct 21, 2026Collaborator·

  ·

• ReloadableDLL: never load a leftover library copy in place of the library

  #2711·ThrudPrimrose opened Oct 18, 2026Collaborator·

  ·
  3

• Add a Python and cuTile backend and lowering pipeline

  #2708·Spielopoly opened Oct 16, 2026Contributor·

  ·
  1

• FFT: N-D and axis-batched transforms, real-input (rfft) family, cached FFTW3/cuFFT/hipFFT plans

  #2693·ThrudPrimrose opened Oct 14, 2026Collaborator·

  ·
  1

• BLAS Scal, Symv, Symm and Syrk nodes, vendor Ger expansions, HIP through hipBLAS

  #2680·ThrudPrimrose opened Oct 11, 2026Collaborator·

  ·
  2

• Per-SDFG symbol registry with explicit facts

  #2673·ThrudPrimrose opened Oct 9, 2026Collaborator·

  ·
  2

• Function regions

  #2658·tbennun opened Oct 6, 2026Collaborator·

  ·
  1

• PyTorch Dynamo Frontend

  no-ci
  #2645·tbennun opened Oct 3, 2026Collaborator·

  ·

• [WIP] New CPU codegen, new GPU codegen and GPU offloading pass (combined)

  no-ci
  #2620·ThrudPrimrose opened Sep 30, 2026Collaborator·

  ·

• Scheduling pass suite for Loop/Map and data reduction based on conditionals

  #2606·tbennun opened Sep 22, 2026Collaborator·

  ·
  42

• OffloadToAcceleratorPass

  #2553·FinnHeckman opened Sep 3, 2026·

  ·
  15

• [New GPU Codegen Prep] Add NestedGPUDeviceMapLowering

  #2534·ThrudPrimrose opened Aug 26, 2026Collaborator·

  ·

• Add static bytes accessed analysis

  #2523·alexanderfluck opened Aug 25, 2026Contributor·

  ·