v1.10.0

Highlights:
- Add experimental JAX automatic differentiation support with jax_kernel(enable_backward=True)
- Add in-place wp.Bvh.rebuild() with CUDA graph support for allocation-free BVH updates
- Improve built-in function call performance from Python by up to 70× through caching
- Add tile programming enhancements: axis-specific reductions, component indexing, wp.tile_full()
- Remove warp.sim module (superseded by Newton library)

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.10.0

v1.9.1

Highlights:

- Fix crash when using radix sort on multiple streams
- Fix memory management issues with shared tiles (double frees, leaks)
- Restore support for older GPU architectures (Maxwell, Pascal, Volta)
  when building with CUDA 12
- Fix TypeError with tuple type hints on Python 3.9/3.10
- Fix empty slice operations arr[i:i] that caused indexing errors

See the full changelog for more details:
https://github.com/NVIDIA/warp/releases/tag/v1.9.1

v1.9.0

Highlights:
- wp.MarchingCubes rewrite in pure Warp, supporting CPU and GPU devices and differentiability
- wp.compile_aot_module() and wp.load_aot_module() to support basic ahead-of-time workflows
- More flexible indexing support for wp.matrix()/wp.vector()/wp.quaternion() types
- Support for IntEnum and IntFlag inside Warp kernels
- Add indexed tile operations: wp.tile_index_load(), wp.tile_index_store(), and wp.tile_index_atomic_add()

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.9.0

v1.9.0rc1

First release candidate (intended for Isaac Lab testing)

v1.8.1

Highlights:

- Deprecate the graph_compatible boolean flag in jax_callable() in favor of the
  new graph_mode argument with GraphMode enum
- Support input-output aliasing in JAX FFI
- Support capturing jax_callable() using Warp via the new graph_mode parameter
- Fix missing cloth-body contact in wp.sim.VBDIntegrator with
  handle_self_contact=False
- Fix compile time regression for kernels using matmul, Cholesky, and FFT
  solvers

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.8.1

v1.8.0

Highlights:
- Add wp.map() to map a function over arrays
- Add support for dynamic control flow in CUDA graphs, see wp.capture_if() and wp.capture_while()
- Add wp.atomic_cas() and wp.atomic_exch() built-ins for atomic compare-and-swap and exchange operations
- Add support for passing tiles to user wp.func functions
- Deprecate the warp.sim module (planned for removal in v1.10)

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.8.0

v1.7.2

Highlights:

- Prevent and document that class inheritance is not supported for wp.struct
  (now throws RuntimeError)
- Fix an issue where arrays stored in structs could be garbage collected without
  updating the struct ctype
- Allow recovering from out-of-memory errors during `wp.Volume` allocation
- Fix 2D tile load when source array and tile have incompatible strides
- Fix `wp.svd2()` with duplicate singular values and improved accuracy
- Warn when an incompatible data type conversion is detected when constructing
  an array using the __cuda_array_interface__

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.7.2

v1.7.2.post1

Highlights:

- Prevent and document that class inheritance is not supported for wp.struct
  (now throws RuntimeError)
- Fix an issue where arrays stored in structs could be garbage collected without
  updating the struct ctype
- Allow recovering from out-of-memory errors during `wp.Volume` allocation
- Fix 2D tile load when source array and tile have incompatible strides
- Fix `wp.svd2()` with duplicate singular values and improved accuracy
- Warn when an incompatible data type conversion is detected when constructing
  an array using the __cuda_array_interface__

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.7.2.post1

v1.7.1

Highlights:
- Include the block dimension in the LTO file hash for the Cholesky solver
- Fix tile loads for small tiles with aligned source memory
- Fix a bug where wp.sim.ModelBuilder adds springs with -1 as vertex indices
- Fix center of mass, inertia computation for mesh shapes

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.7.1

v1.7.0

Highlights:
- Support JAX foreign function interface (FFI).
- Support Python/SASS correlation in Nsight Compute reports.
- Add LTO cache to speed up compilation times for kernels using MathDx-based tile functions.
- Default PTX architecture is now automatically determined by the devices present in the system.
- **Breaking:** Remove CUTLASS dependency and wp.matmul() functionality (including batched version).
- Deprecate wp.select() in favor of wp.where().

See the full changelog for more details: https://github.com/NVIDIA/warp/releases/tag/v1.7.0