The HPVM Compiler Infrastructure

This repository contains the source code and documentation for the HPVM Compiler Infrastructure.

The README briefly describes how to get started with building and installing HPVM. It also provides a benchmark suite to test the compiler infrastructure.

HPVM is currently at version 0.5. For more about what HPVM is, see our website.

Paper

PPoPP'18 paper

Docs

HPVM IR Specification

HPVM-C Language Specification

HPVM Compilation Process

Dependencies

The following components are required to be installed on your machine to build HPVM.

• GCC (>=5.1)
  • In addition, each version of CUDA-nvcc requires GCC to be not newer than a certain version. See here for the support matrix.
• CMake (>=3.17)
• Python (>=3.7) with Pip
• GNU Make (>=3.79)
• OpenCL (>=1.0.0)
• CUDA (>=9.1)

Supported Targets

Supported/tested CPU architectures:

• Intel Xeon E5-2640
• Intel Xeon W-2135
• ARM Cortex A-57

Supported/tested GPU architectures:

• Nvidia Quadro P1000
• Nvidia GeForce GTX 1080

HPVM has not been tested but might work on other CPUs supported by LLVM Backend, and GPUs supported by OpenCL such as Intel, AMD, etc.

Getting source code and building HPVM

Checkout HPVM:

git clone --recursive https://gitlab.engr.illinois.edu/llvm/hpvm-release.git/
cd hpvm-release/hpvm

HPVM needs to be able to find CUDA. If CUDA is installed in your system's $PATH (e.g. if it was installed at the default location), HPVM can find CUDA automatically. Otherwise, some environment variables are required:

• CUDA_TOOLKIT_PATH --- Path to the CUDA toolkit
• CUDA_INCLUDE_PATH --- Path to the CUDA headers
• CUDA_LIB_PATH --- Path to CUDA libraries

hpvm/set_paths.sh can be used for this. Modify the values of these variables in set_paths.sh and source the script:

source set_paths.sh

HPVM installer script can be used to download, configure and build HPVM along with LLVM and Clang.

bash install.sh

Specifically, the HPVM installer downloads LLVM, and Clang, copies HPVM source into llvm/tools and builds the entire tree. It also builds a modified LLVM C-Backend, based on the one maintained by Julia Computing, as a part of HPVM and is currently used to generate OpenCL kernels for GPUs.

In the beginning of the building process, the installer provides users the choice of automatically or manually building HPVM. If HPVM is selected to be built automatically, the installer allows users to type in the number of threads they want to use. The default number of threads used to build HPVM is two.

Alternatively, CMake can be run manually using the following steps in ./hpvm-release/hpvm directory.

mkdir build
cd build
cmake ../llvm [options]

Note that if the installer script was not used, you must manually add build/bin directory to your $PATH variable (as absolute path).

Some common options that can be used with CMake are:

• -DCMAKE_INSTALL_PREFIX=directory --- Specify for directory the full pathname of where you want the HPVM tools and libraries to be installed.

• -DCMAKE_BUILD_TYPE=type --- Valid options for type are Debug, Release, RelWithDebInfo, and MinSizeRel. Default is Debug.

• -DLLVM_ENABLE_ASSERTIONS=On --- Compile with assertion checks enabled (default is Yes for Debug builds, No for all other build types).

In order to manually build and install HPVM, GNU Make can be run using the following in the build directory.

make -j<number of threads>
make install

In the end of the installation process, the installer automatically runs all the regression tests to ensure that the installation is successful. If HPVM is built and installed manually, the tests can be automatically run by executing the following step from the ./hpvm-release/hpvm directory.

bash scripts/automate_tests.sh

With all the aforementioned steps, HPVM should be built, installed, tested and ready to use.

Benchmarks and Tests

We are providing the following HPVM benchmarks:

• Select benchmarks from the Parboil benchmark suite, located under test/benchmarks/parboil.
• An edge detection pipeline benchmark, located under test/benchmarks/pipeline.
• A Camera ISP pipeline, located under test/benchmarks/hpvm-cava, adapted from C code provided from our collaborators at Harvard.

Benchmark descriptions and instructions on how to compile and run them are here.

We are also providing unit tests and regression tests.

Support

All questions can be directed to hpvm-dev@lists.cs.illinois.edu.