CUDA Examples

This repo contains a collection of CUDA examples that were first used for a talk at the Melbourne C++ Meetup.

Listing

1. 00-hello-world.cu - Vector addition on a CPU; the hello world of the parallel computing
2. 01-cuda-hello-world.cu - Vector addition using CUDA
3. 02-cuda-hello-world-faster.cu - Vector addition using CUDA, with a CPU bottleneck removed
4. 03-templates-device-ptr.cu - Using C++ templates to make device pointers safer
5. 04-templates-device-mem.cu - More templates, allowing for scoped device memory
6. 05-thrust-rand-vectors.cu - Using Thrust to add up random numbers

Compiling

All examples can be compiled with nvcc. Only 02-cuda-hello-world-faster.cu requires an additional compiler option --expt-relaxed-constexpr (at least, when compiled on Linux).

A Makefile has been included, so all examples can built using make.

Credits

These are all based on examples found in the wild. 03 and 04, in particular, are based on code from Michael Gopshtein's CppCon talk, CUDA Kernels in C++. And examples 01 and 02 are based on the Vector Addition sample code included in the CUDA Toolkit.

License

This source code has been released into the public domain.