Issues

• Replaying a captured Nanogrid.from_environment_voxels() writes out of bounds in a CUB scan

  bugwarp.fem
  Bug

  NVIDIA/warp#2081 · Filter by author shi-eric opened on Oct 11, 2026

• Out-of-range point_envs values cause out-of-bounds GPU access in sparse grid construction

  bugwarp.fem
  Bug

  NVIDIA/warp#2080 · Filter by author shi-eric opened on Oct 11, 2026 · 1.19.0
  

• tile_matmul reads out when beta is zero on scalar backends

  bugtile
  Bug

  NVIDIA/warp#2079 · Filter by author shi-eric opened on Oct 11, 2026 · 1.19.0
  

• Validate graph coloring balance node colors before native access

  bug
  Bug

  NVIDIA/warp#2078 · Filter by author shi-eric opened on Oct 11, 2026 · 1.19.0
  

• Host-side vector swizzle accepts multi-letter and out-of-range names

  NVIDIA/warp#2076 · Filter by author v0ropaev opened on Oct 10, 2026

• Add wp.bit_count() for integer population count

  feature request
  Enhancement

  NVIDIA/warp#2071 · Filter by author shi-eric opened on Oct 9, 2026 · 1.19.0
  

• Row-packed tiles and mixed scalar types in warp.sparse.bsr_mv

  feature request
  NVIDIA/warp#2069 · Filter by author alecjacobson opened on Oct 9, 2026

• wp.func retains caller frames, causing texture cleanup during CUDA graph capture

  bug
  Bug

  NVIDIA/warp#2068 · Filter by author jcarius-nv opened on Oct 9, 2026 · 1.19.0
  1

  

• Releasing a pinned host array during CUDA graph capture invalidates the capture

  bug
  Bug

  NVIDIA/warp#2067 · Filter by author adenzler-nvidia opened on Oct 9, 2026
  1

  

• Replace runtime rejection of int32/uint32 texture sampling

  language
  Task

  NVIDIA/warp#2065 · Filter by author shi-eric opened on Oct 8, 2026 · 1.19.0
  

• [BUG] fem.interpolate() changes constant vector fields in Nédélec spaces

  bugwarp.fem
  Bug

  NVIDIA/warp#2062 · Filter by author shi-eric opened on Oct 8, 2026 · 1.19.0
  1

  

• [REQ] Support external-source bindings on CPU-only machines

  feature request
  Enhancement

  NVIDIA/warp#2061 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.20.0
  

• [REQ] Support object-oriented syntax for external native types (optional)

  feature request
  Enhancement

  NVIDIA/warp#2060 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.20.0
  

• [REQ] Parse standard-library-bearing C++/CUDA headers with NVRTC

  feature request
  Enhancement

  NVIDIA/warp#2059 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.20.0
  

• [REQ] Declare external C++/CUDA library dependencies for bindings

  feature request
  Enhancement

  NVIDIA/warp#2058 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.20.0
  

• [REQ] Register external C++/CUDA types at runtime

  feature request
  Enhancement

  NVIDIA/warp#2057 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.20.0
  

• [REQ] Register external C++/CUDA functions at runtime

  feature request
  Enhancement

  NVIDIA/warp#2056 · Filter by author daedalus5 opened on Oct 8, 2026 · 1.19.0
  

• Half-precision svd3 returns NaNs for finite diagonal matrices

  bug
  Bug

  NVIDIA/warp#2055 · Filter by author shi-eric opened on Oct 8, 2026
  1

• Half-precision svd2 and svd3 produce NaN gradients for singular-value-only losses

  bug
  Bug

  NVIDIA/warp#2054 · Filter by author shi-eric opened on Oct 8, 2026

• Freeing a wp.Texture during a CUDA graph capture fails the capture (error 900)

  bug
  Bug

  NVIDIA/warp#2053 · Filter by author nvlukasz opened on Oct 8, 2026
  1

  

• Grouped CUDA LBVH builds shallower group subtrees when constructed with more groups

  bugspatial acceleration
  Bug

  NVIDIA/warp#2052 · Filter by author shi-eric opened on Oct 7, 2026 · 1.19.0
  

• wp.rand* only partially have Pass-by-reference semantics for its state parameter.

  bug
  NVIDIA/warp#2050 · Filter by author YouJiacheng opened on Oct 7, 2026
  1

• Wrong adjoint when a sibling dynamic loop reassigns a variable from an earlier loop inside an enclosing dynamic loop

  NVIDIA/warp#2048 · Filter by author jaredmejia opened on Oct 6, 2026

• iter_reverse() yields a spurious iteration for empty strided ranges (wrong adjoints for grid-stride loops)

  NVIDIA/warp#2046 · Filter by author jaredmejia opened on Oct 6, 2026

• Keep CPU fiber tile semantics consistent with CUDA

  tile
  Task

  NVIDIA/warp#2044 · Filter by author shi-eric opened on Oct 6, 2026 · 1.19.0