To run all the experiments on GraphFlow (mini version): $ sh test_GraphFlow.sh Test for the CPU/C++ contraction of GraphFlow: $ g++ test_GraphFlow.cpp -o test_GraphFlow_cpu; ./test_GraphFlow_cpu 10 20 Test for the GPU/CUDA contraction of GraphFlow: $ nvcc test_GraphFlow.cu -o test_GraphFlow_cuda; ./test_GraphFlow_cuda 10 20 Mini version of the deep learning framework is in GraphFlow_gpu/ directory: * Entity: Entity.h * Vector: Vector.h * Matrix: Matrix.h * 3-dimensional tensor: Tensor3D.h * 4-dimensional tensor: Tensor4D.h * CPU/C++ contraction operator: RisiContraction_18.h * GPU/CUDA contraction operator: RisiContraction_18_gpu.h
HyTruongSon/risi_contraction
Testing contraction operations in C++/CUDA GraphFlow deep learning framework vs PyTorch
C++