Nvhpc

Introduction

The NVIDIA HPC SDK is a comprehensive suite of compilers, libraries and tools essential to maximizing developer productivity and the performance and portability of HPC applications. The NVIDIA HPC SDK C, C++, and Fortran compilers support GPU acceleration of HPC modeling and simulation applications with standard C++ and Fortran, OpenACC directives, and CUDA. GPU-accelerated math libraries maximize performance on common HPC algorithms, and optimized communications libraries enable standards-based multi-GPU and scalable systems programming. Performance profiling and debugging tools simplify porting and optimization of HPC applications, and containerization tools enable easy deployment on-premises or in the cloud. For more information, please check: NGC: https://ngc.nvidia.com/catalog/containers/nvidia:nvhpc

Versions

  • 20.11

  • 20.7

  • 20.9

  • 21.5

  • 21.9

Commands

  • nvc

  • nvc++

  • nvfortran

  • nvcc

  • pgcc

  • pgc++

  • pgfortran

  • cuda-gdb

  • ncu

  • nv-nsight-cu-cli

  • nvprof

  • nsight-sys

  • nsys

Module

You can load the modules by:

module load ngc
module load nvhpc

Example job

Warning

Using #!/bin/sh -l as shebang in the slurm job script will cause the failure of some biocontainer modules. Please use #!/bin/bash instead.

To run nvhpc on our clusters:

#!/bin/bash
#SBATCH -A myallocation     # Allocation name
#SBATCH -t 1:00:00
#SBATCH -N 1
#SBATCH -n 1
#SBATCH --job-name=nvhpc
#SBATCH --mail-type=FAIL,BEGIN,END
#SBATCH --error=%x-%J-%u.err
#SBATCH --output=%x-%J-%u.out

module --force purge
ml ngc nvhpc