Skip to content
main
Switch branches/tags
14 branches 10 tags
Code

Latest commit

@MasterJH5574
MasterJH5574 [BugFix][TIR] Fix construction of new IterVars in CacheRead/Write for…
ae285c6 Mar 26, 2022
[BugFix][TIR] Fix construction of new IterVars in CacheRead/Write for… 
… non-int32 dtypes (#10795)

_This PR is a follow-up effort of #10789, which enables the `int64` support for TIR schedule primitive Cache-Read and Cache-Write._

Prior to this PR, the IterVars of the generated cache stage block are always `int32`-typed, which might conflict with the dtypes of the domains of the IterVars.

In this PR, the dtype of new IterVars are constructed according to the data types of their domains, and thereby the possible conflicts are resolved. Meanwhile the data types of the read/write regions of the cache stage blocks are also constructed according to correct data types.
ae285c6

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
.github
 
 
3rdparty
 
 
apps
 
 
cmake
 
 
conda
 
 
docker
 
 
docs
 
 
gallery
 
 
golang
 
 
include/tvm
 
 
jenkins
 
 
jvm
 
 
licenses
 
 
nnvm
 
 
python
 
 
rust
 
 
src
 
 
tests
 
 
vta
 
 
web
 
 
.asf.yaml
 
 
.clang-format
 
 
.gitignore
 
 
.gitmodules
 
 
.pre-commit-config.yaml
 
 
CMakeLists.txt
 
 
CONTRIBUTORS.md
 
 
Jenkinsfile
 
 
KEYS
 
 
LICENSE
 
 
Makefile
 
 
NEWS.md
 
 
NOTICE
 
 
README.md
 
 
conftest.py
 
 
mypy.ini
 
 
pyproject.toml
 
 
version.py
 
 
Open Deep Learning Compiler Stack License Getting Started Contribute to TVM Acknowledgement

README.md

Open Deep Learning Compiler Stack

Documentation | Contributors | Community | Release Notes

Build Status WinMacBuild

Apache TVM is a compiler stack for deep learning systems. It is designed to close the gap between the productivity-focused deep learning frameworks, and the performance- and efficiency-focused hardware backends. TVM works with deep learning frameworks to provide end to end compilation to different backends.

License

TVM is licensed under the Apache-2.0 license.

Getting Started

Check out the TVM Documentation site for installation instructions, tutorials, examples, and more. The Getting Started with TVM tutorial is a great place to start.

Contribute to TVM

TVM adopts apache committer model, we aim to create an open source project that is maintained and owned by the community. Check out the Contributor Guide.

Acknowledgement

We learned a lot from the following projects when building TVM.

  • Halide: Part of TVM's TIR and arithmetic simplification module originates from Halide. We also learned and adapted some part of lowering pipeline from Halide.
  • Loopy: use of integer set analysis and its loop transformation primitives.
  • Theano: the design inspiration of symbolic scan operator for recurrence.

About

Open deep learning compiler stack for cpu, gpu and specialized accelerators

tvm.apache.org/

Topics

javascript machine-learning performance deep-learning metal compiler gpu vulkan opencl tensor spirv rocm tvm

Resources

Readme

License

Apache-2.0 License

Code of conduct

Code of conduct

Stars

7.9k stars

Watchers

377 watching

Forks

2.5k forks

Releases 10

Apache TVM v0.8 Release Note Latest
Nov 24, 2021
+ 9 releases

Packages

No packages published

Contributors 674

+ 663 contributors

Languages