I managed to get the n-body simulation producing the correct result, and hit the performance target too. Thanks, so much for the CUDA course and cert!
$ make -j8 nbody-opt && build/nbody-opt mkdir -p build /usr/local/cuda/bin/nvcc -arch compute_60 -o build/nbody-opt nbody-opt.cu Simulator is calculating positions correctly. 4096 Bodies: average 35.231 Billion Interactions / second
(on a GTX 1080)
I managed to do the integration on the GPU as well. I have one question, however: What is the license on our solutions since they’re based off Nvidia examples?