/libwb

My (private) copy of libwb. The purpose of this fork to solve the "machine problem"s of the coursera class: Heterogenous Paralell Programming

Primary LanguageC++University of Illinois/NCSA Open Source LicenseNCSA

libwb

Compiling and Running on Linux and Windows

by Tran Minh Quan

This is a tutorial explains how to compile and run your Machine Problems (MPs) offline without separating on building libwb.

Caution: If you don't have NVIDIA GPUs (CUDA Capable GPUs) on your local machine, you cannot run the executable binaries.

First, regardless your platform is, please install CUDA 5.5 and Cmake 2.8 (http://www.cmake.org/) , then set the path appropriate to these things (linux).

Check out the source codes (only skeleton codes) for MPs as following

https://github.com/hvcl/hetero13

git clone https://github.com/abduld/libwb

1. If you are under Linux environment, you should use gcc lower than 4.7 (mine is 4.4.7). Ortherwise, it will not be compatible with nvcc

cd libwb
ls
mkdir build
cd build/
cmake ..
make -j4
./MP0

2. If you are under Windows environment

Open Cmake Gui:

Where is the source code: {libwb}/
Where to build the binary: {libwb}/build

image

Press Configure, Yes and choose your compiler (in this case Visual Studio 10 (32 bit) or Visual Studio 10 Win64 (64 bit), then press Finish

image

image

Press Configure one more time and generate

image

Open your generated folder and Double click on libwb.sln

image

Right click to MP0 and click "set as startup project"

Press Ctrl F5

Whenever you do the MPs, change the MP accordingly.

Best regards,

P/s1: Sorry about the name of project, it should be hetero14 but I forget that we are already in the new year. My appologize :-P

P/s2: If you are using MAC, please consider reading this link and modify your CMakeLists.txt https://class.coursera.org/hetero-002/forum/thread?thread_id=83