Issues

• MathDx calls from a @wp.func fail to link in CUDA 13 debug builds

  NVIDIA/warp#2249 · Filter by author alecjacobson opened on Nov 6, 2026

• CPU-block lanes use stale floating-point control state (FTZ, rounding)

  NVIDIA/warp#2248 · Filter by author alecjacobson opened on Nov 6, 2026

• array_scan that works inside conditional bodies and foreign captures (caller-provided temp storage, or a capture query)

  NVIDIA/warp#2247 · Filter by author jaredmejia opened on Nov 5, 2026
  

• Let tile_cholesky_inplace() skip zeroing the opposite triangle

  feature requesttile
  NVIDIA/warp#2246 · Filter by author adenzler-nvidia opened on Nov 5, 2026
  

• float16 atan2 loses gradients after coordinate scaling

  NVIDIA/warp#2244 · Filter by author DiadkoShmek opened on Nov 5, 2026

• [BUG] warp.optim.linear solvers are slow on CPU and ignore use_cuda_graph=True

  bug
  NVIDIA/warp#2242 · Filter by author christophercrouzet opened on Nov 5, 2026
  

• Avoid redundant FP64 diagonal work in native tile Cholesky

  bugtile
  NVIDIA/warp#2239 · Filter by author alecjacobson opened on Nov 5, 2026
  

• Add wp.tile_ldlt with Bunch-Kaufman pivoting for symmetric indefinite matrices

  feature request
  Enhancement

  NVIDIA/warp#2238 · Filter by author alecjacobson opened on Nov 4, 2026 · 1.20.0
  

• Add wp.tile_lu with cuSOLVERDx acceleration

  feature requesttile
  Enhancement

  NVIDIA/warp#2237 · Filter by author alecjacobson opened on Nov 4, 2026 · 1.20.0
  

• smoothstep interval powers corrupt representable gradients

  NVIDIA/warp#2234 · Filter by author DiadkoShmek opened on Nov 4, 2026

• Float64 Grid3D PicQuadrature lookup receives a float32 distance bound

  bugwarp.fem
  Bug

  NVIDIA/warp#2232 · Filter by author alecjacobson opened on Nov 4, 2026 · 1.20.0
  

• wp.utils.array_cast reads freed or wrong source data when ndim or element shape differ

  NVIDIA/warp#2230 · Filter by author udsy19 opened on Nov 3, 2026

• Atomic mapped sparse updates fail for empty matrices

  bug
  NVIDIA/warp#2228 · Filter by author JaimeFine opened on Nov 3, 2026

• wp.norm_l1, norm_huber, norm_pseudo_huber and smooth_normalize fail to compile for float16/float64 vectors

  NVIDIA/warp#2224 · Filter by author 0z5a opened on Nov 3, 2026

• Gradient of wp.mod / % with respect to the divisor is always zero for floats

  NVIDIA/warp#2223 · Filter by author 0z5a opened on Nov 3, 2026

• Iterative solvers with check_every=0 run all maxiter iterations without CUDA graphs

  bug
  Bug

  NVIDIA/warp#2222 · Filter by author alecjacobson opened on Nov 3, 2026 · 1.20.0
  1

  

• Float64 Nanogrid lookup mixes distance precision during PicQuadrature binning

  NVIDIA/warp#2220 · Filter by author alecjacobson opened on Nov 3, 2026

• Float64 Cube B-spline FEM assembly uses float32 node coordinates and gradients

  NVIDIA/warp#2218 · Filter by author alecjacobson opened on Nov 3, 2026

• Clang CUDA backend (wp.config.llvm_cuda = True) can't load kernels on GPUs older than sm_75

  NVIDIA/warp#2215 · Filter by author 0z5a opened on Nov 2, 2026

• In-kernel wp.zeros() copies a zero-filled temporary, doubling CUDA stack use

  bug
  Bug

  NVIDIA/warp#2214 · Filter by author adenzler-nvidia opened on Nov 2, 2026 · 1.20.0
  

• wp.utils.runlength_encode() ignores strides and reads past the end of values

  NVIDIA/warp#2212 · Filter by author bodapatisaikrishna opened on Nov 2, 2026

• [REQ] Coalesced multiple-RHS application of FSAI factors

  NVIDIA/warp#2206 · Filter by author alecjacobson opened on Nov 2, 2026
  

• [REQ] Reusable GMRES solve actions with fresh device residual diagnostics

  feature request
  NVIDIA/warp#2205 · Filter by author alecjacobson opened on Nov 2, 2026
  

• [REQ] Reusable multiple-RHS CG solves for one shared sparse operator

  feature request
  NVIDIA/warp#2204 · Filter by author alecjacobson opened on Nov 2, 2026
  

• [BUG] CPU graph replay uses stale meshes and volumes

  apicbug
  Bug

  NVIDIA/warp#2202 · Filter by author christophercrouzet opened on Nov 1, 2026