##PyTorch for GPGPU-Sim using CUDNN Contributor: Jonathan Lew, University of British Columbia, Dept. of Electrical and Computer Engineering
Welcome to PyTorch for GPGPU-Sim!
This repo is based on the popular PyTorch framework, release v0.4.1, the latest release as of Aug 17, 2018. It can be found at https://github.com/pytorch/pytorch
Note: Only Linux install has been tested. Instructions for other operating systems are from the pytorch repo and are left here for your convenience.
Install
- NVIDIA CUDA 7.5 or above
- NVIDIA cuDNN v6.x or above
If you want to build on Windows, Visual Studio 2017 and NVTX are also needed.
git clone --recursive [this repo]
cd pytorch-gpgpu-simOn Linux
python setup.py installUse the following if you encounter root permission error:
python setup.py install --userOn macOS
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py installOn Windows
set "VS150COMNTOOLS=C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Auxiliary\Build"
set CMAKE_GENERATOR=Visual Studio 15 2017 Win64
set DISTUTILS_USE_SDK=1
REM The following line is needed for Python 2.7, but the support for it is very experimental.
set MSSdk=1
REM As for CUDA 8, VS2015 Update 2 or up is required to build PyTorch. Use the following two lines.
set "PREBUILD_COMMAND=%VS140COMNTOOLS%\..\..\VC\vcvarsall.bat"
set PREBUILD_COMMAND_ARGS=x64
call "%VS150COMNTOOLS%\vcvarsall.bat" x64 -vcvars_ver=14.11
python setup.py install-
Install GPGPU-Sim with CUDNN and PyTorch support at:
-
Follow the README to set the environment variables.
-
export PYTORCH_BIN=/path/to/libcudnn.so
Three pointers to get you started: