Physis: A High-Level Domain-Specific Framework for Stencil Computations

This is under active development, and is not yet ready for production use. The features listed here should work, but you would likely encounter issues that do not work as expected. Email the author if you are interested in more details.

Overview

Physis is a framework for stencil computations that is designed for a variety of parallel computing systems with a particular focus on programmable GPUs. The primary goals are high productivity and high performance. Physis DSL is a small set of custom programming constructs, and allows for very concise and portable implementations of common stencil computations. A single Physis program runs on x86 CPUs, NVIDIA GPUs, and even clusters of them with no platform-specific code.

This software consists of a DSL translator and runtime layer for each supported platform. The translator automatically generates platform-specific source code from Physis code, which is then compiled by a platform-native compiler to generate final executable code. The runtime component is a thin software layer that performs application-independent common tasks, such as management of GPU devices and network connections.

Features

• C-based embedded DSL consisting of declarative intrinsics
• Platform portability
• Virtual shared memory
• No explicit parallel programming required
• Tested with the fourth fastest supercomputer

News & Updates

• 19 Aug 2013 - Version 0.5 available.
• 25 Jul 2012 - Version 0.4 available.
• 24 Apr 2012 - Version 0.3 available.
• 1 Jan 2012 - Version 0.2 available.
• 22 Nov 2011 - Version 0.1 available.

Supported Platforms

• Linux and Mac OS X running on x86 CPUs
• CUDA-capable NVIDIA GPUs
• Clusters of machines with MPI

Usage

External Dependencies

Physis depends on the following external software:

• GNU GCC C/C++ compiler
• ROSE compiler infrastructure
• Boost C++ libraries
• CMake
• Lua compiler and runtime libraries

In addition, the following platform-specific tools and libraries are required when using the respective platform:

• NVIDIA CUDA toolkit and SDK for using NVIDIA GPUs (6.5 tested)
• MPI for multi-node parallel execution (OpenMPI v1.4.2 tested)

Installing Physis

See docs/install.md.

Writing Physis Code

See docs/programming.md. Some example code is available at the examples directory.

Compiling Physis Programs

See docs/compilation.md.

FAQ

See docs/faq.md.

Version History

The corresponding source of each version is attached with a Git tag with the version number.

• Version 0.5 (19 Aug 2013)
  • User-defined point type.
  • Only REFERENCE and CUDA targets are updated. MPI and MPI-CUDA targets are not updated in this version.
• Version 0.4 (25 Jul 2012)
  • Single GPU tuning. Near hand-tuned performance with 7-pt diffusion.
• Version 0.3 (24 Apr 2012)
  • ROSE usage changed to imporve robustness of AST translation.
• Version 0.2 (1 Jan 2012)
  • Optimization passes added.
• Version 0.1 (22 Nov 2011)
  • The first pre-release version.

Publications

• Naoya Maruyama, Tatsuo Nomura, Kento Sato, and Satoshi Matsuoka, "Physis: An Implicitly Parallel Programming Model for Stencil Computations on Large-Scale GPU-Accelerated Supercomputers," Proceedings of the 2011 ACM/IEEE Conference on Supercomputing (SC'11), pp. 1--12, Seattle, WA, USA, Nov 2011.

License (BSD License)

Copyright (c) 2011-2012, Naoya Maruyama

All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

• Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
• Neither the name of RIKEN AICS nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.