{"id":64015,"date":"2013-03-14T17:04:00","date_gmt":"2013-03-14T21:04:00","guid":{"rendered":"http:\/\/www.bu.edu\/tech\/?page_id=64015"},"modified":"2023-07-21T13:58:16","modified_gmt":"2023-07-21T17:58:16","slug":"cuda-fortran","status":"publish","type":"page","link":"https:\/\/www.bu.edu\/tech\/support\/research\/software-and-programming\/gpu-computing\/cuda-fortran\/","title":{"rendered":"Programming for GPUs using CUDA in Fortran"},"content":{"rendered":"

CUDA is a parallel programming model and software environment developed by NVIDIA. It provides programmers with a set of instructions that enable GPU acceleration for data-parallel computations. The computing performance of many applications can be dramatically increased by using CUDA directly or by linking to GPU-accelerated libraries.<\/p>\n

Setting up your environment<\/h2>\n

Your environment should by default be all set to take advantage of CUDA Fortran using the Portland Group Fortran compiler pgfortran<\/span><\/code>. To use pgfortran<\/span><\/code> you will need to load the nvidia-hpc\/2023-23.5 module using the command<\/span><\/p>\n

scc1%<\/span> module load nvidia-hpc\/2023-23.5<\/span><\/span><\/code><\/pre>\n
Compiling and running interactively a simple CUDA program using Portland Group CUDA Fortran<\/h2>\n
Run man pgfortran<\/span><\/code> for usage instructions.<\/p>\n
There are two CUDA Fortran free-format source file suffixes; .cuf and .CUF. The .CUF files require preprocessing.<\/p>\n
As a test, you can download the CUDA Fortran matrix multiply example matmul.cuf<\/a> and transfer it to the directory where you are working on the SCC.<\/p>\n
You should do your compiling of CUDA Fortran programs on one of our nodes with GPUs, not on the login nodes<\/b>. You can get access to a GPU-equipped node by running the command below.<\/p>\n
scc1%<\/span> qrsh<\/span> -l gpus=1<\/span><\/code><\/pre>\n
After that, to compile this CUDA Fortran program, run:<\/p>\n
scc-je2%<\/span> pgfortran<\/span> -fast -o matmul matmul.cuf<\/span><\/code><\/pre>\n
To run a CUDA program interactively, you then type in the name of the program at the command prompt:<\/p>\n
gpunode%<\/span> matmul<\/span><\/code><\/pre>\n
Submit a CUDA program Batch Job<\/h2>\n
The following line shows how to submit the matmul<\/code> executable to run in batch mode on a single CPU with access to a single GPU:<\/p>\n
scc1%<\/span> qsub -l gpus=1<\/span> -b y matmul<\/span><\/code><\/pre>\n
CUDA Libraries<\/h2>\n
Several scientific libraries that make use of CUDA are available:<\/p>\n
where the \u2013l gpus=<\/span>#<\/span><\/code> option indicates the number of GPUs. To learn about all options that could be used for submitting a job, please visit the running jobs page<\/a>.<\/p>\n