#GPU Graph Applications in CUDA This suite contains Graph Applications used in the following IISWC paper:
Q. Xu, H. Jeon, M. Annavaram (University of Southern California), Graph processing on GPUs: Where are the bottlenecks?, In Proceedings of the IEEE International Symposium on Workload Characterization (IISWC), Oct. 2014.
Please cite if your use of this repository results in a publication.
#System Requirements
- Linux® (tested with CentOS release 5.9)
- CUDA 4.0 & CUDA 5.5
- NVIDIA's CUDA development libraries and tools
- Cusp v0.4.0
- Boost development libraries
- Intel's Threading Building Blocks library
#Installation and Building
- Put the Graph Application folder under NVIDIA_GPU_Computing_SDK4/C/ directory
- Add the following lines in NVIDIA_GPU_Computing_SDK4/C/common/common.mk
ifeq ($(USEBOOST),1) LIB += -lboost_iostreams-mt -lboost_program_options-mt endif
ifeq ($(USETBB),1) LIB += -ltbb endif
- Step into each folder and type make.
#Running
- Run the applications using the sample commands in the run file.
- To run GCL, MIS, PR on GPGPUSim, you may want to patch the GPGPUSim with CUDA5.5 capability. A detailed instruction is provided here: http://iss.ices.utexas.edu/?p=projects/galois/lonestargpu/gpgpusim Other benchmarks work well with CUDA4.0.
#Notice BFS, MST, SP, SSSP are not included here. Please refer to other existing benchmark suites (Rodinia, Lonestar).