NVHPC
- Robert Schade
- Axel Keller
- xinwu
Description
C, C++ and Fortran compilers included with the NVIDIA HPC SDK (previously: PGI)
More information
- Homepage: https://developer.nvidia.com/hpc-sdk/
Available Versions of NVHPC
Version | Module | Available on |
|---|---|---|
23.1-CUDA-12.0.0 | compiler/NVHPC/23.1-CUDA-12.0.0 | Noctua 1 |
22.11 | compiler/NVHPC/22.11 | Noctua 2 |
21.11 | compiler/NVHPC/21.11 | Noctua 1, Noctua 2 |
This table is generated automatically. If you need other versions please click pc2-support@uni-paderborn.de.
Usage Hints for NVHPC
If you need support in using this software or example job scipts please contact pc2-support@uni-paderborn.de.
OpenMP GPU Offloading
NVIDIA HPC SDK (NVHPC) supports OpenMP GPU Offloading. The following is an example SLURM jobscript using one NVIDIA A100 GPU on Noctua 2. After loading the NVHPC module file, it compiles a simple SAXPY code written in OpenMP GPU offloading and then executes it to measure the GPU memory bandwidth.
#!/usr/bin/env bash
#SBATCH --job-name=saxpy_gpu
#SBATCH --nodes=1
#SBATCH --partition=gpu
#SBATCH --gres=gpu:a100:1
#SBATCH --time=00:05:00
#
# load NVIDIA HPC SDK
#
module reset
module load compiler/NVHPC/21.11
# compilation
nvc -fast -O4 -mp=gpu -o saxpy_gpu.x saxpy_gpu.c
# execution for measurement of GPU memory bandwidth
./saxpy_gpu.x $((256*2**20))