ansform_layout - Applies a layout transformation to a buffer as specified by an IndexMap.Several TIR schedule primitives and transforms including (abridged):.MLF - Support for multiple modules in a single MLF artifact.Testing via Hexagon SDK simulator and on device via Snapdragon-based HDK boards and phones.

AOT - Support for the C++ runtime (with llvm and c targets only) and support for host-driven AOT in the C runtime.

New parameters - runtime=, executor=,.

ARM cascading scheduler for Arm Ethos(TM)-U NPUs.

v0.9.0 is the first release on the new quarterly release schedule and includes many highlights, such as: The TVM community has worked since the v0.8 release to deliver many exciting features and improvements.

#12607 - enhance compute_at and reverse_compute_at primitive to choose possible position.

#12623 - Expose MMA-related PTX builtins.

#12652 - Handle axis_separators during FlattenBuffer.

#12717 - Check consumer in-bound and covered in reverse_compute_inline.

#12515 - Add more strict checking in imm construction and folding.

#11639 - Tuple-reduction scheduling support.

#12750 - Add schedule primitive PadEinsum.

#12827 - Support pattern matching argmax/argmin generated by TOPI.

#12797 - Construct the inverse in SuggestIndexMap.

#12720 - Implement API for padded layout transformations.

A non-truncated summary can be found here: #12979 TIR Note that this list is not comprehensive of all PRs and discussions since v0.9. Please visit the full listing of commits for a complete view: v0.9.0.v0.10.0rc0. Create LLVM scope class for use with LLVM libraries (#83) ( 22d1d11) Name mangling in IRModules (#84) ( 831d702)Īsynchronous stage in software pipeline (#80) ( aecb219) TVMScript Metaprogramming (#79) ( ffbf686)Īdd Target Pre-processing RFC (#71) ( 78423c5)

Add Commit Message Guideline (#88) ( e8a2d8b) Issue Triage Workflow RFC (#93) ( 3345cc1) These RFCs have been merged in apache/tvm-rfcs since the last release. Improved schedule primitives for ARM v8-m ISAĪnd many other general improvements to code quality, TVMScript, and more! Please visit the full listing of commits for a complete view: v0.9.0.v0.10.0rc0.INT8 Intrinsics for TensorCores for CUDA!.Stabilized and polished user-interfaces (e.g.Software pipelining and padding for irregular shapes for auto tensorization.The TVM community has worked since the v0.9 release to deliver the following new exciting improvments!

Tvmc download zip file