Open Deep Learning Compiler Stack

Documentation | Contributors | Community | Release Notes

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. Checkout the tvm stack homepage for more information.

安装

基本安装步骤参考TVM官方文档

本fork repo额外的依赖

1. 需要安装python yaml包。例如使用pip install pyyaml安装。
2. 需要编译并安装yaml-cpp库。
3. 下载并按说明编译这个LLVM repo，注意需要切换到release_70分支！并在编译TVM时，在build/config.cmake中修改set(USE_LLVM /path/to/your/llvm/bin/llvm-config)，令它指向刚刚编译的LLVM。
4. 下载并编译安装SystemC 2.3.3。建议在编译SystemC时使用C++14标准（cmake -DCMAKE_CXX_STANDARD=14 ...），否则可能出现加载libsystemc.so时符号找不到的错误。

运行NNPU demo

• 首先将tvm/nnpu/python添加到PYTHONPATH环境变量。
• tvm/nnpu/demos目录下有各个算子的演示代码，例如执行python test_conv.1.py执行卷积算子演示。使用--sim S0/SC参数选择功能级/行为级仿真器,例如python test_conv.2.py --sim SC。
• 配置文件tvm/nnpu/config/nnpu_config.yaml包含NNPU后端硬件属性的配置参数，也可以将其复制到tvm/nnpu/目录下。

License

© Contributors Licensed under an Apache-2.0 license.

Contribute to TVM

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

Acknowledgement

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

• Halide: TVM uses HalideIR as data structure for arithmetic simplification and low level lowering. We also learnt 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.