Retour haut de page

Home > Products > HMPP Workbench

Rapidly develop GPU-accelerated applications

HMPP Workbench

HMPP Workbench, a directive-based compiler for hybrid computing

 

Based on C and FORTRAN directives, HMPP offers a high level abstraction for hybrid programming that fully leverages the computing power of stream processors without the complexity associated with GPU programming. 

HMPP compiler integrates powerful data-parallel backends for NVIDIA CUDA and OpenCL that drastically reduce development time. The HMPP runtime ensures application deployment on multi-GPU systems. Software assets are kept independent from both hardware platforms and commercial software. While preserving portability and hardware interoperability, HMPP increases application performance and development productivity.

Download Datasheet
  • A COMPILER WITH POWERFUL BACK-END GENERATORS

    With the HMPP back-end code generators, instantaneously prototype and evaluate the performance of hardware accelerated critical functions.

    The code generators are specifically designed to extract the most of data parallelism from your C and Fortran kernels and translate them into NVIDIA CUDA or OpenCL. HMPP Workbench includes a C and Fortran compiler, back-end code generators and a runtime that seamlessly integrate in your environment and make use of the CUDA / OpenCL development tools and drivers.
  • A HIGH LEVEL ABSTRACTION FOR MANYCORE PROGRAMMING

    Based on a set of OpenMP™-like directives that preserve legacy codes, HMPP fully leverages the performance offered by most of today’s stream processors and vector units. HMPP offers a standardized interface between your scientific algorithm and fast evolving target code by insulating hardware specific implementation of functions from your legacy code.
    Complementary to OpenMP and MPI, HMPP lets you develop parallel hybrid applications that mix the best of today’s available parallel tools.
  • DYNAMIC APPLICATION SCALING

    By providing different target versions of computations that are offloaded to the available hardware compute units, an HMPP application dynamically adapts its execution to multi-GPUs systems and platform configuration. This guarantees the scalability and interoperability of your application.
  • SUPPORTED PLATFORMS AND COMPILERS
    GPUs
    All NVIDIA Tesla and AMD ATI FireStream
     
    Compilers
    Intel 11.1 and above
    GNU gcc 4.1 and above / GNU gfortran 4.3 and above
    Open64 4.2
    PGI 10.0
    SUN 12.1
     
     

    Operating systems

    Any x86_64 kernel 2.6 Linux distribution with libc coming with g++ 4.x and above.

    HMPP has been validated with some of the below Linux distributions:

    • Debian 5.8 and above
    • RedHat Entreprise Linux 5.x and above
    • OpenSuse 11.x and above
    • SLES 11.0
    • Ubuntu 9.04
     
    - Windows
     
Back to top