This repo contains source code for the GPU computing workshop.
The example source files are based on the CUDA SDK Samples:
https://github.com/NVIDIA/cuda-samples
To build the source codes (the .cu files in each folder) on Midway3, do
module load cuda/11.5 openmpi/4.1.2+gcc-10.2.0
and then go to each folder and do
cd ex1-scale
nvcc -O2 -arch=sm_70 scale.cu -o vec_scale
where sm_70 means that you want the executable vec_scale to run on NVIDIA V100 GPUs.
To run the executables, you need a GPU node with at least a GPU:
sinteractive -A [your-allocation] --reservation=[reservation-name-if-needed] -p gpu --gres=gpu:1
Once on the compute node, load the modules again (if they are not loaded) before running the executables
module load cuda/11.5 openmpi/4.1.2+gcc-10.2.0
cd ex1-scale
./vec_scale