NVIDIA Performance Libraries

The NVIDIA Performance Libraries (NVPL) are a collection of high performance mathematical libraries optimized for the NVIDIA Grace Arm v9 architecture.

These CPU-only libraries have no dependencies on CUDA or CTK, and are drop in replacements for standard C and Fortran mathematical APIs allowing HPC applications to achieve maximum performance on the Grace platform.

Libraries Documentation

Installation

System Support

  • Architecture: aarch64-linux

  • Platform: Arm SBSA

CPUs Supported

  • NVIDIA Grace (Arm-v9.0a Neoverse-V2)

  • AWS Graviton 3/3e (Arm-v8.4a Neoverse-V1)

  • AWS Graviton 2 (Arm-v8.2a Neoverse-N1)

  • Ampere Altra (Arm-v8.2a Neoverse-N1)

  • Any CPU with Arm-v8.2a+ architecture

OS Support

  • AmazonLinux: 2, 2023

  • Debian: 12

  • Fedora: 37. 38, 39

  • RHEL: RHEL8, RHEL9

  • SLES: SLES15

  • Ubuntu: 20.04, 22.04, 23.10

  • Generally any Linux OS with support for aarch64

Compiler Support

Languages

  • C: All libraries

  • C++: All libraries via C interfaces

  • Fortran: GFortran ABI - Selected libraries

    • GFortran ABI

    • NVPL BLAS, LAPACK, and ScaLAPACK provide lp64 and ilp64 integer ABIs

    • See individual libraries documentation for further details

OpenMP

All libraries support the following OpenMP runtime libraries. See individual libraries documentation for details and API extensions supporting nested parallelism.

  • GCC: libgomp.so

  • Clang: libomp.so

  • NVHPC: libnvomp.so

MPI

NVPL provides standard BLACS interfaces for the following MPI distributions. See the NVPL ScaLAPACK Documentation for details.