Changelog

[1.8.0] - 2025-07-01

Added

• Addwp.map() to map a function over arrays and add math operators for Warp arrays (docs,#694).
• Add support for dynamic control flow in CUDA graphs, seewp.capture_if() andwp.capture_while() (docs,#597).
• Addwp.capture_debug_dot_print() to write a DOT file describing the structure of a captured CUDA graph (#746).
• Add theDevice.sm_count property to get the number of streaming multiprocessors on a CUDA device (#584).
• Addwp.block_dim() to query the number of threads in the current block inside a kernel (#695).
• Addwp.atomic_cas() andwp.atomic_exch() built-ins for atomic compare-and-swap and exchange operations (#767).
• Add support for profiling GPU runtime module compilation using the globalwp.config.compile_time_trace setting or the module-level"compile_time_trace" option. When used, JSON files in the Trace Event format will be written in the kernel cache, which can be opened in a viewer likechrome://tracing/ (docs,#609).
• Add support for returning multiple values from native functions likewp.svd3() andwp.quat_to_axis_angle() (#503).
• Add support for passing tiles to userwp.func functions (#682).
• Addwp.tile_squeeze() to remove axes of length one (#662).
• Addwp.tile_reshape() to reshape a tile (#663).
• Addwp.tile_astype() to return a new tile with the same data but different data type. (#683).
• Add support for in-place tile add and subtract operations (#518).
• Add support for in-place tile-component addition and subtraction (#659).
• Add support for 2D solves usingwp.tile_cholesky_solve() (#773).
• Addwp.tile_scan_inclusive() andwp.tile_scan_exclusive() for performing inclusive and exclusive scans over tiles (#731).
• Support attribute indexing for quaternions on the right-hand side of expressions (#625).
• Addwp.transform_compose() andwp.transform_decompose() for converting between transforms and 4x4 matrices with 3D scale information (#576).
• Add variouswp.transform syntax operations for loading and storing (#710).
• Add theas_spheres parameter toUsdRenderer.render_points() in order to choose whether to render the points as USD spheres using a point instancer or as simple USD points (#634).
• Add support for animating visibility of objects in the USD renderer (#598).
• Addwp.sim.VBDIntegrator.rebuild_bvh() to rebuild the BVH used for detecting self-contacts.
• Add damping termswp.sim.VBDIntegrator collisions, with strength is controlled byModel.soft_contact_kd.
• Improve consistency of thewp.fem.lookup() operator across geometries and add filtering parameters (#618).
• Add two examples demonstrating shape optimization usingwarp.fem:fem/example_elastic_shape_optimization.py andfem/example_darcy_ls_optimization.py (#698).
• Add apy.typed marker file (per PEP 561) to the package to formally support static type checking by downstream users (#780).

Removed

• Removewp.mlp() (deprecated in v1.6.0). Use tile primitives instead.
• Removewp.autograd.plot_kernel_jacobians() (deprecated in v1.4.0). Usewp.autograd.jacobian_plot() instead.
• Remove thelength andowner keyword arguments fromwp.array() constructor (deprecated in v1.6.0). Use theshape anddeleter keywords instead.
• Remove thekernel keyword argument fromwp.autograd.jacobian() andwp.autograd.jacobian_fd() (deprecated in v1.6.0). Use thefunction keyword argument instead.
• Remove theoutputs keyword argument fromwp.autograd.jacobian_plot() (deprecated in v1.6.0).

Changed

• Deprecate thewarp.sim module (planned for removal in v1.10). It will be superseded by the upcoming Newton library, a separate package with a new API. Migrating will require code changes; a future guide will be provided (current draft). See the GitHub announcement for details (#735).
• Deprecate thewp.matrix(pos, quat, scale) built-in function. Usewp.transform_compose() instead (#576).
• Improve support for tuples in kernels (#506).
• Return a constant value fromlen() where possible.
• Rename the internal functionwp.types.type_length() towp.types.type_size().
• Renamewp.tile_cholesky_solve() input parameters to align with its docstring (#726).
• Changewp.tile_upper_solve() andwp.tile_lower_solve() to use libmathdx 0.2.1 TRSM solver (#773).
• Skip adjoint compilation forwp.tile_matmul() ifenable_backward is disabled (#644).
• Allow tile reductions to work with non-scalar tile types (#771).
• Permit data-type preservation withpreserve_type=True when tiling a value across the block withwp.Tile() (#772).
• Makewp.sparse.bsr_[set_]from_triplets differentiable with respect to the input triplet values (#760).
• Expose newwarp.fem operators:node_count,node_index,element_coordinates,element_closest_point.
• Changewp.sim.VBDIntegrator rigid-body-contact handling to use only the shape's friction coefficient, rather than averaging the shape's and the cloth's coefficients.
• Limit usage of thewp.assign_copy() hidden built-in to the kernel scope.
• Describe the distinction betweeninputs andoutputs arguments in theKernel documentation.
• Reduce the overhead ofwp.launch() by avoiding costly native API calls (#774).
• Improve error reporting when calling@wp.func-decorated functions from the Python scope (#521).

Fixed

• Fix missing documentation for geometric structs (#674).
• Fix the type annotations in various tile functions (#714).
• Fix incorrect stride initialization in tiles returned from functions taking transposed tiles as input (#722).
• Fix adjoint generation for user functions that return a tile (#749).
• Fix tile-based solvers failing to accept and return transposed tiles (#768).
• Fix theFormal parameter space overflowed error duringwp.sim.VBDIntegrator kernel compilation for the backward pass in CUDA 11 Warp builds. This was resolved by decoupling collision and elasticity evaluations into separate kernels, increasing parallelism and speeding up the solver (#442).
• Fix an issue with graph coloring on an empty graph (#509).
• Fix an integer overflow bug in the native graph coloring module (#718).
• FixUsdRenderer.render_points() not supporting multiple colors (#634).
• Fix an inconsistency in thewp.fem module regarding the orientation of 2D geometry side normals (#629).
• Fix premature unloading of CUDA modules used in JAX FFI graph captures (#782).

Changelog

[1.7.2] - 2025-05-31

Added

• Add missing adjoint method for tileassign operations (#680).
• Add documentation for the fact that+= and-= invokewp.atomic_add() andwp.atomic_sub(), respectively (#505).
• Add apublications list of academic and research projects leveraging Warp (#686).

Changed

• Prevent and document that class inheritance is not supported forwp.struct (now throwsRuntimeError) (#656).
• Warn when an incompatible data type conversion is detected when constructing an array using the__cuda_array_interface__ (#624,#670).
• Relax the exact version requirement inomni.warp towardsomni.warp.core (#702).
• Rename the "Kernel Reference" documentation page to "Built-Ins Reference", with each built-in now having annotations to denote whether they are accessible only from the kernel scope or also from the Python runtime scope (#532).

Fixed

• Fix an issue where arrays stored in structs could be garbage collected without updating the struct ctype (#720).
• Fix an issue with preserving the base class of nested struct attributes (#574).
• Allow recovering from out-of-memory errors duringwp.Volume allocation (#611).
• Fix 2D tile load when source array and tile have incompatible strides (#688).
• Fix compilation errors withwp.tile_atomic_add() (#681).
• Fixwp.svd2() with duplicate singular values and improved accuracy (#679).
• FixOpenGLRenderer.update_shape_instance() not having color buffers created for the shape instances.
• Fix text rendering inwp.render.OpenGLRenderer (#704).
• Fix assembly of rigid body inertia inModelBuilder.collapse_fixed_joints() (#631).
• FixUsdRenderer.render_points() erroring out when passed 4 points or less (#708).
• Fixwp.atomic_*() built-ins not working with some types (#733).
• Fix garbage-collection issues with JAX FFI callbacks (#711).

Changelog

[1.7.1] - 2025-04-30

Added

• Add example of a distributed Jacobi solver usingmpi4py inwarp/examples/distributed/example_jacobi_mpi.py (#475).

Changed

• Improverepr() for Warp types, including addingrepr() forwp.array.
• Change the USD renderer to useframesPerSecond for time sampling instead oftimeCodesPerSecond to avoid playback speed issues in some viewers (#617).
• Model.rigid_contact_tids are now -1 at non-active contact indices which allows to retrieve the vertex index of a mesh collision, seetest_collision.py (#623).
• Improve handling of deprecated JAX features (#613).

Fixed

• Fix a code generation bug involving return statements in Warp kernels, which could result in some threads in Warp being skipped when processed on the GPU (#594).
• Fix constructingDeformedGeometry fromwp.fem.Trimesh3D geometries (#614).
• Fixlookup operator forwp.fem.Trimesh3D (#618).
• Include the block dimension in the LTO file hash for the Cholesky solver (#639).
• Fix tile loads for small tiles with aligned source memory (#622).
• Fix length/shape matching for vectors and matrices from the Python scope.
• Fix thedtype parameter missing forwp.quaternion().
• Fix invaliddtype comparison when using thewp.matrix()/wp.vector()/wp.quaternion() constructors with literal values and an explicitdtype argument (#651).
• Fix incorrect thread index lookup for the backward pass ofwp.sim.collide() (#459).
• Fix a bug wherewp.sim.ModelBuilder adds springs with -1 as vertex indices (#621).
• Fix center of mass, inertia computation for mesh shapes (#251).
• Fix computation of body center of mass to account for shape orientation (#648).
• Fixshow_joints not working withwp.sim.render.SimRenderer set to render to USD (#510).
• Fix the jitter for theOgnParticlesFromMesh node not being computed correctly.
• Fix documentation ofatol andrtol arguments towp.autograd.gradcheck() andwp.autograd.gradcheck_tape() (#508).

Changelog

[1.7.0] - 2025-03-30

Added

• Support JAX foreign function interface (FFI) (docs,#511).
• Support Python/SASS correlation in Nsight Compute reports by emitting#line directives in CUDA-C code. This setting is controlled bywp.config.line_directives and isTrue by default. (docs,#437)
• Supportvec4f grid construction inwp.Volume.allocate_by_tiles().
• Add 2D SVDwp.svd2() (#436).
• Addwp.randu() for randomuint32 generation.
• Add matrix construction functionswp.matrix_from_cols() andwp.matrix_from_rows() (#278).
• Addwp.transform_from_matrix() to obtain a transform from a 4x4 matrix (#211).
• Addwp.where() to select between two arguments conditionally using a more intuitive argument order (cond,value_if_true,value_if_false) (#469).
• Addwp.get_mempool_used_mem_current() andwp.get_mempool_used_mem_high() to query the respective current and high-water mark memory pool allocator usage (#446 ).
• AddStream.is_complete andEvent.is_complete properties to query completion status (#435).
• Support timing events inside of CUDA graphs (#556).
• Add LTO cache to speed up compilation times for kernels using MathDx-based tile functions. Usewp.clear_lto_cache() to clear the LTO cache (#507).
• Add example demonstrating gradient checkpointing for fluid optimization inwarp/examples/optim/example_fluid_checkpoint.py.
• Add a hinge-angle-based bending force towp.sim.VBDIntegrator.
• Add an example to show mesh sampling using a CDF (#476).

Changed

• Breaking: Remove CUTLASS dependency andwp.matmul() functionality (including batched version). Users should use tile primitives for matrix multiplication operations instead.
• Deprecate constructing a matrix from vectors usingwp.matrix().
• Deprecatewp.select() in favor ofwp.where(). Users should update their code to usewp.where(cond, value_if_true, value_if_false) instead ofwp.select(cond, value_if_false, value_if_true).
• wp.sim.Control no longer has amodel attribute (#487).
• wp.sim.Control.reset() is deprecated and now only zeros-out the controls (previously restored controls to initialmodel state). Usewp.sim.Control.clear() instead.
• Vector/matrix/quaternion component assignment operations (e.g.,v[0] = x) now compile and run faster in the backward pass. Note: For correct gradient computation, each component should only be assigned once.
• @wp.kernel has now an optionalmodule argument that allows passing awp.context.Module to the kernel, or, if set to"unique" let Warp create a new unique module just for this kernel. The default behavior to use the current module is unchanged.
• Default PTX architecture is now automatically determined by the devices present in the system, ensuring optimal compatibility and performance (#537).
• Structs now have a trivial default constructor, allowing forwp.tile_reduce() on tiles with struct data types.
• Extendwp.tile_broadcast() to support broadcasting to 1D, 3D, and 4D shapes (in addition to existing 2D support).
• wp.fem.integrate() andwp.fem.interpolate() may now perform parallel evaluation of quadrature points within elements.
• wp.fem.interpolate() can now build Jacobian sparse matrices of interpolated functions with respect to a trial field.
• Multiplewp.sparse routines (bsr_set_from_triplets,bsr_assign,bsr_axpy,bsr_mm) now accept amasked flag to discard any non-zero not already present in the destination matrix.
• wp.sparse.bsr_assign() no longer requires source and destination block shapes to evenly divide each other.
• Extendwp.expect_near() to support all vectors and quaternions.
• Extendwp.quat_from_matrix() to support 4x4 matrices.
• Update theOgnClothSimulate node to use the VBD integrator (#512).
• Remove theglobalScale parameter from theOgnClothSimulate node.

Fixed

• Fix an out-of-bounds access bug caused by an unbalanced BVH tree (#536).
• Fix an error of incorrectly adding the offset to -1 elements inedge_indices when adding a ModelBuilder to another (#557).

Changelog

[1.6.2] - 2025-03-07

Changed

• Update project license fromNVIDIA Software License toApache License, Version 2.0 (seeLICENSE.md).

Changelog

[1.6.1] - 2025-03-03

Added

• Documentwp.Launch objects (docs,#428).
• Document how overwriting previously computed results can lead to incorrect gradients (docs,#525).

Fixed

• Fix unaligned loads with offset 2D tiles inwp.tile_load().
• Fix FP64 accuracy of thread-level matrix-matrix multiplications (#489).
• Fixwp.array() not initializing from arrays defining a CUDA array interface when the target device is CPU (#523).
• Fixwp.Launch objects not storing and replaying adjoint kernel launches (#449).
• Fixwp.config.verify_autograd_array_access failing to detect overwrites in generic Warp functions (#493).
• Fix an error on Windows when closing anOpenGLRenderer app (#488).
• Fix per-vertex colors not being correctly written out to USD meshes when a constant color is being passed (#480).
• Fix an error in capturing thewp.sim.VBDIntegrator with CUDA graphs whenhandle_self_contact is enabled (#441).
• Fix an error of AABB computation inwp.collide.TriMeshCollisionDetector.
• Fix URDF-imported planar joints not being set with the intendedtarget_ke,target_kd, andmode parameters (#454).
• FixModelBuilder.add_builder() to use correct offsets forModelBuilder.joint_parent andModelBuilder.joint_child (#432)
• Fix underallocation of contact points for box–sphere and box–capsule collisions.
• Fixwp.randi() documentation to show correct output range of[-2^31, 2^31).

Changelog

[1.6.0] - 2025-02-03

Added

• Add preview of Tile Cholesky factorization and solve APIs throughwp.tile_cholesky(),tile_cholesky_solve()
  andtile_diag_add() (preview APIs are subject to change).
• Support for loading tiles from arrays whose shapes are not multiples of the tile dimensions.
  Out-of-bounds reads will be zero-filled and out-of-bounds writes will be skipped.
• Support for higher-dimensional (up to 4D) tile shapes and memory operations.
• Add intersection-free self-contact support inwp.sim.VDBIntegrator by passinghandle_self_contact=True.
  Seewarp/examples/sim/example_cloth_self_contact.py for a usage example.
• Add functionswp.norm_l1(),wp.norm_l2(),wp.norm_huber(),wp.norm_pseudo_huber(), andwp.smooth_normalize()
  for vector types to a newwp.math module.
• wp.sim.SemiImplicitIntegrator andwp.sim.FeatherstoneIntegrator now have an optionalfriction_smoothing
  constructor argument (defaults to 1.0) that controls softness of the friction norm computation.
• Supportassert statements in kernels (docs).
  Assertions can only be triggered in"debug" mode (GH-366).
• Support CUDA IPC on Linux. Call theipc_handle() method to get an IPC handle for awp.Event or awp.array,
  and callwp.from_ipc_handle() orwp.event_from_ipc_handle() in another process to open the handle
  (docs).
• Add per-module option to disable fused floating point operations, usewp.set_module_options({"fuse_fp": False})
  (GH-379).
• Add per-module option to add CUDA-C line information for profiling, usewp.set_module_options({"lineinfo": True}).
• Support operator overloading forwp.struct objects by definingwp.func functions
  (GH-392).
• Add built-in functionwp.len() to retrieve the number of elements for vectors, quaternions, matrices, and arrays
  (GH-389).
• Addwarp/examples/optim/example_softbody_properties.py as an optimization example for soft-body properties
  (GH-419).
• Addwarp/examples/tile/example_tile_walker.py, which reworks the existingexample_walker.py
  to use Warp's tile API for matrix multiplication.
• Addwarp/examples/tile/example_tile_nbody.py as an example of an N-body simulation using Warp tile primitives.

Changed

• Breaking: Changewp.tile_load() andwp.tile_store() indexing behavior so that indices are now specified in
  terms ofarray elements instead oftile multiples.
• Breaking: Tile operations now takeshape andoffset parameters as tuples,
  e.g.:wp.tile_load(array, shape=(m,n), offset=(i,j)).
• Breaking: Change exception types and error messages thrown by tile functions for improved consistency.
• Add an implicit tile synchronization whenever a shared memory tile's data is reinitialized (e.g. in dynamic loops).
  This could result in lower performance.
• wp.Bvh constructor now supports various construction algorithms via theconstructor argument, including
  "sah" (Surface Area Heuristics),"median", and"lbvh" (docs)
• Improve the query efficiency ofwp.Bvh andwp.Mesh.
• Improve memory consumption, compilation and runtime performance when using in-place vector/matrix assignments in
  kernels that haveenable_backward set toFalse (GH-332).
• Vector/matrix/quaternion component+= and-= operations compile and run faster in the backward pass
  (GH-332).
• Name files in the kernel cache according to their directory. Previously, all files began with
  module_codegen (GH-431).
• Avoid recompilation of modules when changingblock_dim.
• wp.autograd.gradcheck_tape() now has additional optional argumentsreverse_launches andskip_to_launch_index.
• wp.autograd.gradcheck(),wp.autograd.jacobian(), andwp.autograd.jacobian_fd() now also accept
  arbitrary Python functions that have Warp arrays as inputs and outputs.
• update_vbo_transforms kernel launches in the OpenGL renderer are no longer recorded onto the tape.
• Skip emitting backward functions/kernels in the generated C++/CUDA code whenenable_backward is set toFalse.
• Emit deprecation warnings for the use of theowner andlength keywords in thewp.array initializer.
• Emit deprecation warnings for the use ofwp.mlp(),wp.matmul(), andwp.batched_matmul().
  Use tile primitives instead.

Fixed

• Fix unintended modification of non-Warp arrays during the backward pass (GH-394).
• Fix so thatwp.Tape.zero() zeroes gradients passed via thegrads parameter inwp.Tape.backward()
  (GH-407).
• Fix errors during graph capture caused by module unloading (GH-401).
• Fix potential memory corruption errors when allocating arrays with strides (GH-404).
• Fixwp.array() not respecting the targetdtype andshape when the given data is an another array with a CUDA interface
  (GH-363).
• Negative constants evaluate to compile-time constants (GH-403)
• FixImportError exception being thrown during interpreter shutdown on Windows when using the OpenGL renderer
  (GH-412).
• Fix the OpenGL renderer not working when multiple instances exist at the same time (GH-385).
• FixAttributeError crash in the OpenGL renderer when moving the camera (GH-426).
• Fix the OpenGL renderer not correctly displaying duplicate capsule, cone, and cylinder shapes
  (GH-388).
• Fix the overriding ofwp.sim.ModelBuilder default parameters (GH-429).
• Fix indexing ofwp.tile_extract() when the block dimension is smaller than the tile size.
• Fix scale and rotation issues with the rock geometry used in the granular collision SDF example
  (GH-409).
• Fix autodiff Jacobian computation inwp.autograd.jacobian() where in some cases gradients were not zeroed-out properly.
• Fix plotting issues inwp.autograd.jacobian_plot().
• Fix thelen() operator returning the total size of a matrix instead of its first dimension.
• Fix gradient instability in rigid-body contact handling forwp.sim.SemiImplicitIntegrator and
  wp.sim.FeatherstoneIntegrator (GH-349).
• Fix overload resolution of generic Warp functions with default arguments.
• Fix rendering of arrows with differentup_axis,color inOpenGLRenderer (GH-448).

Changelog

[1.5.1] - 2025-01-02

Added

• Add PyTorch basics and custom operators notebooks to thenotebooks directory.
• Update PyTorch interop docs to include section on custom operators
  (docs).

Fixed

• warp.sim: Fix a bug in which the color-balancing algorithm was not updating the colorings.
• Fix custom colors being not being updated when rendering meshes with static topology in OpenGL
  (GH-343).
• Fixwp.launch_tiled() not returning aLaunch object when passedrecord_cmd=True.
• Fix default arguments not being resolved forwp.func when called from Python's runtime
  (GH-386).
• Array overwrite tracking: Fix issue with not marking arrays passed towp.atomic_add(),wp.atomic_sub(),
  wp.atomic_max(), orwp.atomic_min() as being written to (GH-378).
• Fix for occasional failure to update.meta files into Warp kernel cache on Windows.
• Fix the OpenGL renderer not being able to run without a CUDA device available
  (GH-344).
• Fix incorrect CUDA driver function versions (GH-402).

Changelog

[1.5.0] - 2024-12-02

Added

• Support for cooperative tile-based primitives using cuBLASDx and cuFFTDx, please see the tile
  documentation for details.
• Expose areversed() built-in for iterators (GH-311).
• Support for saving Volumes into.nvdb files with thesave_to_nvdb method.
• warp.fem: Addwp.fem.Trimesh3D andwp.fem.Quadmesh3D geometry types for 3D surfaces with newexample_distortion_energy example.
• warp.fem: Add"add" option towp.fem.integrate() for accumulating integration result to existing output.
• warp.fem: Add"assembly" option towp.fem.integrate() for selecting between more memory-efficient or more
  computationally efficient integration algorithms.
• warp.fem: Add Nédélec (first kind) and Raviart-Thomas vector-valued function spaces
  providing conforming discretization ofcurl anddiv operators, respectively.
• warp.sim: Add a graph coloring module that supports converting trimesh into a vertex graph and applying coloring.
  Thewp.sim.ModelBuilder now includes methods to color particles for use withwp.sim.VBDIntegrator(),
  users should callbuilder.color() before finalizing assets.
• warp.sim: Add support for a per-particle radius for soft-body triangle contact using thewp.sim.Model.particle_radius
  array (docs), replacing the previous
  hard-coded value of 0.01 (GH-329).
• Add aparticle_radius parameter towp.sim.ModelBuilder.add_cloth_mesh() andwp.sim.ModelBuilder.add_cloth_grid()
  to set a uniform radius for the added particles.
• Documentwp.array attributes (GH-364).
• Document time-to-compile tradeoffs when using vector component assignment statements in kernels.
• Add introductory Jupyter notebooks to thenotebooks directory.

Changed

• Drop support for Python 3.7; Python 3.8 is now the minimum-supported version.
• Promote thewp.Int,wp.Float, andwp.Scalar generic annotation types to the public API.
• warp.fem: Simplify querying neighboring cell quantities when integrating on sides using new
  wp.fem.cells(),wp.fem.to_inner_cell(),wp.fem.to_outer_cell() operators.
• Show an error message when the type returned by a function differs from its annotation, which would have led to the compilation stage failing.
• Clarify thatwp.randn() samples a normal distribution of mean 0 and variance 1.
• Raise error when passing more than 32 variadic argument to thewp.printf() built-in.

Fixed

• Fixplace setting of paddle backend.
• warp.fem: Fix tri-cubic shape functions on quadrilateral meshes.
• warp.fem: Fix caching of integrand kernels when changing code-generation options.
• Fixwp.expect_neq() overloads missing for scalar types.
• Fix an error when awp.kernel or awp.func object is annotated to return aNone value.
• Fix error when reading multi-volume, BLOSC-compressed.nvdb files.
• Fixwp.printf() erroring out when no variadic arguments are passed (GH-333).
• Fix memory access issues in soft-rigid contact collisions (GH-362).
• Fix gradient propagation for in-place addition/subtraction operations on custom vector-type arrays.
• Fix the OpenGL renderer's window not closing when clicking the X button.
• Fix the OpenGL renderer's camera snapping to a different direction from the initial camera's orientation when first looking around.
• Fix custom colors being ignored when rendering meshes in OpenGL (GH-343).
• Fix topology updates not being supported by the the OpenGL renderer.

Changelog

[1.4.2] - 2024-11-13

Changed

• Make the output ofwp.print() in backward kernels consistent for all supported data types.

Fixed

• Fix to relax the integer types expected when indexing arrays (regression in1.3.0).
• Fix printing vector and matrix adjoints in backward kernels.
• Fix kernel compile error when printing structs.
• Fix an incorrect user function being sometimes resolved when multiple overloads are available with array parameters with differentdtype values.
• Fix error being raised when static and dynamic for-loops are written in sequence with the same iteration variable names (GH-331).
• Fix an issue with theTexture Write node, used in the Mandelbrot Omniverse sample, sometimes erroring out in multi-GPU environments.
• Code generation of in-place multiplication and division operations (regression introduced ina69d061)(GH-342).