このページの更新は終了しました。

最新の情報はTSUBAME3.0計算サービスのWebページをご覧ください。

TSUBAME2.5からTSUBAME3.0へのデータ移行方法の資料はこちら

What would be changed by maintenance in 2016 spring

Changes summary at 2016 spring.

1. OS/Libraly update
  OS is no change.
  Library changes following:
      You can use the following update application.
      Along with this upgrade, change has come out to the store software, of command. 
      As follows.

  CUDA 7.5 (/usr/apps.sp3/cuda/7.5/)
  OpenMPI
     1.6.5 (gcc4.3.4) ((/usr/apps.sp3/mpi/openmpi/1.6.5/g4.3.4/)
     1.6.5 (pgi16.1) (/usr/apps.sp3/mpi/openmpi/1.6.5/p16.1/)
     1.6.5 (i2013.1.046) (/usr/apps.sp3/mpi/openmpi/1.6.5/i2013.1.046/) ←default(no change)
     1.8.2 ((gcc4.3.4) (/usr/apps.sp3/mpi/openmpi/1.8.2/g4.3.4_cuda7.5/)
     1.8.2 (pgi16.1) (/usr/apps.sp3/mpi/openmpi/1.8.2/p16.1_cuda7.5/)
     1.8.2 (i2013.1.046) (/usr/apps.sp3/mpi/openmpi/1.8.2/i2013.1.046_cuda7.5/)
   Mvapich2
     2.0rc1 (gcc4.3.4) (/usr/apps.sp3/mpi/mvapich2/2.0rc1/g4.3.4_cuda7.5/)
     2.0rc1 (pgi16.1) (/usr/apps.sp3/mpi/mvapich2/2.0rc1/p16.1_cuda7.5/)
     2.0rc1 (i2013.1.046) (/usr/apps.sp3/mpi/mvapich2/2.0rc1/i2013.1.046_cuda7.5/)
   mpich2
     3.1 (gcc4.3.4) (/usr/apps.sp3/mpi/mpich2/3.1/g4.3.4/)
     3.1 (pgi16.1) (/usr/apps.sp3/mpi/mpich2/3.1/p16.1/)
     3.1 (i2013.1.046) (/usr/apps.sp3/mpi/mpich2/3.1/i2013.1.046/)

 

2.Default version of the application
It is updated as follows.
When you use an old version, please appoint the pass.

     ANSYS                     16.0  -> 16.2 
     Patran                    2013  -> 2014.1 
     MD Nastran                2013  -> 2014.1 
     MARC                    2013.1  -> 2015 
     LS-DYNA                 R7.1.1  -> R8.0.0 
     Gaussian           09 Rev D.01  -> 09 Rev E.01 
     AMBER             14 update 12  -> 14 update 13 
     Mathematica              10.0.2 -> 10.3 ->  (10.4 is not use in TSUBAME)
     MATLAB                   R2015a -> R2016a
     AVS/Express                8.2  -> 8.3 
     PGI C/Fortran Compiler     15.7 ->  16.1

3.Change Queue
    Reduction of compute nodes hos occurred.

   Queue  number of nodes
   S      338 → 300
   S96     41 → 30
   G      480 → 434
   V      430 → 384 (max)
   U      240 → 194(max)
   L128    not use
   L128F   not use
   L256     8 (no GPU)
   L512     2 (no GPU)
   H/X                   420  → 380 
   (Hmax       )     (475) → (420)
   win/interactive   no change