Geintra
Departamento de electronica Universidad de Alcala

Líneas de investigación

Accede a información sobre la estructura de la actividad investigadora de Geintra.

Más información

Trabaja con nosotros

Accede a nuestra oferta actual de becas, tesis doctorales, contratos y trabajos fin de carrera.

Más información

Contacta con el grupo

Si desea contactar con nosotros, puede usar varios medios.

Más información
Inicio › Ayuda

    GEINTRA Cluster: use of NVIDIA GPUs

    Mar, 21/01/2014 - 11:57am — javier.macias

    gc007, being also a node in the cluster infrastructure and thus able to run general jobs, has support for NVIDIA GPUS.

    The general characteristics of this node are: 

    •      SUPERSERVER SYS-2026GT-TRF-FM409 SUPERMICRO
    •      2 x INTEL DP WESTMERE 4C E562' 2.4g 12m 5.86gt
    •      6 x DDR3 1333 4GB ECC REGISTERED
    •      2 x SATA3 WD5000BPKT WESTERN DIGITAL 500GB 7200 16MB
    •      Current GPU configuration includes 4 TESLA UNITS M2090
     I've also installed basic CUDA development support (CUDA 5.5, the latest version as of January 2014). The general details for compiling and linking are (from the installation log):
     
    ---------------------------------------------------------------------------
    Driver:   Installed
    Toolkit:  Installed in /usr/share/geintra/apps/cuda/cuda-5.5
    Samples:  Installed in /usr/share/geintra/apps/cuda/cuda-5.5-samples
     
    * Include path is /usr/share/geintra/apps/cuda/cuda-5.5/include
     
    * Please make sure your LD_LIBRARY_PATH
    *   for 32-bit Linux distributions includes /usr/share/geintra/apps/cuda/cuda-5.5/lib
    *   for 64-bit Linux distributions includes /usr/share/geintra/apps/cuda/cuda-5.5/lib64:/usr/share/geintra/apps/cuda/cuda-5.5/lib
     
    Please see CUDA_Getting_Started_Linux.pdf in /usr/share/geintra/apps/cuda/cuda-5.5/doc/pdf for detailed information on setting up CUDA.
    ---------------------------------------------------------------------------
     
    To ease further updates in the CUDA libraries, I suggest you to use the following paths (unless you depend on a given SDK version):
    • Include path /usr/share/geintra/apps/cuda/cuda/include
    • Link path:
      • For 32-bit Linux distributions /usr/share/geintra/apps/cuda/cuda/lib
      • For 64-bit Linux distributions /usr/share/geintra/apps/cuda/cuda/lib64:/usr/share/geintra/apps/cuda/cuda/lib
    The tests have been run and work properly.
     
    I've also installed the Tesla deployment library, including:
    • Compilation support for the NVML library (same paths than above)
    • Sample, at  /usr/share/geintra/apps/cuda/nvml-samples
    • nvidia-healthmon at /usr/share/geintra/apps/cuda/nvidia-healthmon

    If you need additional supporting tools and/or libraries, please tell me at macias@depeca.uah.es

     

    Geintra © 2008-2024

    Diseño web por Hazhistoria