CCMS

Center for Computational Materials Science
Institute for Materials Research,
Tohoku University

  1. Large-Scale Parallel Computing Server (Cray XC50-LC) manuals
  2. Accelerator Server (Cray CS-Storm) manuals
  3. Parallel computing and Informatics Server manuals


Large-Scale Parallel Computing Server (Cray XC50-LC) manuals

User Guide

Check the following PDF data.

Compiler manuals

Library manual

Cray LibSci

  1. LibSci outline
    login to "super.sc.imr.tohoku.ac.jp", and run the following command.
    $ man intro_libsci
  2. BLAS(Basic Linear Algebra Subroutines)
    login to "super.sc.imr.tohoku.ac.jp", and run the following command.
    BLAS1: $ man blas1
    BLAS2: $ man blas2
    BLAS3: $ man blas3
  3. LAPACK(Linear Algebra routines)
  4. ScaLAPACK(parallel Linear Algebra routines)
    ScaLAPACK outline: $ man intro_scalapack
  5. BLACS
    login to "super.sc.imr.tohoku.ac.jp", and run the following command.
    $ man blacs
  6. IRT(Iterative Refinement Toolkit)
    Cray IRT outline and functions: $ man intro_irt
  7. SuperLU DIST(sparse solvers)
    SuperLU users guide
    online documents

Accelerator Server (Cray CS-Storm) manuals

User Guide

Check the following PDF data.

Compiler manuals

Library manuals

GPU-Accelerated CUDA Libraries

  1. CUDA API References

Parallel computing and Informatics Server manuals

Parallel computing and informatics server consists of the following servers.

  1. Cloud service server (calculation server)
  2. Cloud service server (virtual server)
  3. Visualization server

Purpose

  1. Cloud service server (calculation server)
  2. Users can submit jobs through PBS job scheduling software from the front-end nodes. It is mainly used for running non-parallelized programs or small jobs which run with less than 2 nodes. The main memory size is 567 GiB per node. It is less than Large-scale parallel computing server and Accelerator server, but the CPU frequency is better than them. If you run small jobs, it may be efficient to use this server. Also, users can run several non-parallelized jobs in one job simultaneously.


  3. Cloud service server (virtual server)
  4. Users can use to occupy this server (32 cores per node, 36 cores per user, and available period is up to 4 weeks) by reserving from reservation portal site. This server provides data integration, analysis, and visualization functions to promote research by materials informatics. Users can use server which has 32 cores up to 4 weeks. So, users can run long time job in this server.

    Reservation portal site:
    https://cmsportal.sc.imr.tohoku.ac.jp/reservation_portal/login


  5. Visualization server
  6. Users can use application programs to visualize simulation results.


User Guide

Check the following PDF data.