TVM Monthly - November 2023

Announcement
#1

Note: This monthly report contains main branch only.

As discussed by the TVM PMC, our goal is to provide a monthly summary of the project so users and developers can get a better understanding of the goings on of the TVM community.

Feedback and suggestions are welcomed so that we can further improve these updates.

Community

  • #16172 - Yixin Dong → Reviewer
  • #16164 - Qiang Zhang → Committer
  • #16162 - Shuai Yuan → Committers
  • #16166 - Bohan Hou → PMC
  • #16165 - Ruihang Lai → PMC

RFCs

A new RPC about SYCL Backend was proposed. In short, Add a new backend language——SYCL, enhancing TVM’s compatibility and portability across different types of accelerators.

We continue to improve Relay, TIR, Frontend and other Runtimes. Note again: This monthly report contains main branch only.

Adreno

  • #15991 - [CI]Enhancements to Adreno specific CI utils
  • #15786 - [TOPI]Add conv2d transpose nchw texture schedule

Metal

  • #16033 - [Codegen]Disable cross-function call in Metal codegen

cuda & cutlass & tensorrt

  • #16061 - [CUDA] Add an option for profiling cuda kernels

Frontend

  • #16159 - [KERAS] Fix bug concat convert for NCHW
  • #15914 - [TFLite]Support quantized SQUARE

Relay

  • #16143 - [Pytorch] Add support for aten::scaled_dot_product_attention
  • #16167 - [BugFix] [Pytorch] Fix missing .dtype
  • #16123 - [Pytorch] Add support for aten::linalg_vector_norm
  • #16106 - [TOPI]Add conv2d NHWC hybrid schedule for arm_cpu
  • #16151 - conv3d depthwise bug fix
  • #16131 - [Pytorch] Add support for aten::unflatten
  • #16105 - [Pytorch] Add support for aten::bitwise_and
  • #16079 - [Pytorch] Add support for aten::swapaxes operator

TOPI

  • #16052 - Enhance topi.nn.matmul
  • #16080 - Reduce code redundancy in conv2d weights transformation

Runtime

  • #16133 - Parallel-for with threading backend
  • #16066 - Support clear global memory allocators
  • #16030 - Introduce TVM_MODULE_VTABLE Macros

TIR

  • #16070 - [Transform] Introduce new InjectPermutedLayout pass
  • #16083 - Enhance Python Type Annotations for TIR Expr
  • #16073 - Support more mma intrinsics and get_mma_intrin_group utility
  • #16076 - Enhance Python Type Annotations for TIR stmt
  • #16074 - Fix the thread binding iter_var dtype in Bind primitive
  • #16063 - Fix pass RenewDefs error in gather/take case
  • #16027 - Fix software pipeline with dynamic loop extent

TVMScript

  • #16041 - Fix mismatched dtype of IterVar in T.thread_binding

BugFix

  • #16091 - [Fix] Fix topi.rms_norm with float32 upscale
  • #16081 - [Fix] Broken Windows Build with LLVM
  • #16051 - [Fix][TIR] Fix dtype issues for match_buffer and ramp node
  • #14655 - [VTA] Fix FSIM compile error on macOS
  • #16021 - [FFI] Typo fix of IncRef to DecRef

CI

  • #16110 - Refactor unittest folder
  • #16055 - Fix broken links about Jenkins
  • #16062 - Use LLVM 17 for tests on ci_arm

Docs

  • #16152 - [DOC] Add v0.14.0 docs to site
  • #16127 - Revert “[#15157][Rust][Doc] Re-enable the Rust documentation build (#15213)”
  • #16097 - Add missing backtick to contribute/code_guide.rst
  • #16089 - Fix error on linting by adding --rev argument
  • #16024 - Update release_process.rst about version number modification

Misc

  • #15968 - Add missing backtick
  • #16034 - [Packaging] Include BYOC dynamic libraries into wheel
  • #16087 - Add _ffi_api.py under script folder
  • #16039 - [Target] Support obtain l2 cache size from target
  • #16065 - [Pylint] fix pylint issues from test_random to test_tedd
  • #16031 - [TRT] fix outdated module building method in tensorrt
  • #16032 - [CMake] Use llvm-config to locate Findzstd.cmake
  • #16023 - [Pylint] fix pylint issues for thrust&tflite_runtime&util
  • #15998 - [Codegen] Add shuffle for cuda and metal