mtcrawshaw / dynamic-sparse-graph

This is an implementation of the paper "Dynamic Sparse Graph for Efficient Deep Learning" using CUDA.

dynamic-sparse-graph

This is an implementation of the paper "Dynamic Sparse Graph for Efficient Deep Learning" using CUDA.

Edit Makefile so that NVCCFLAGS contains the correct value of --gpu-architecture and --gpu-code corresponding to your GPU. A list of the correct values for each generation can be found here: https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#gpu-feature-list.
Run 'make'.

./mlp_driver

This is an implementation of the paper "Dynamic Sparse Graph for Efficient Deep Learning" using CUDA.

Language:Cuda 89.7%Language:C 7.9%Language:C++ 1.6%Language:Makefile 0.8%