Intel® Xeon Phi™ Coprocessor enabled applications on Stampede

BelindaLiviero · ‎11-18-2014

This is intended to be an informational post -- I received the following announcement as I am on a TACC mailing list, wanted to pass the word around about the fact that several applications are readily available from their Stampede cluster -- more below (what follows is their announcement in entirety)

-----------------------------------------------------------------------------------------------------------------------------------------------

New Xeon Phi Applications on Stampede

From: Jesse Snead
TACC is pleased to announce the availability of additional Xeon Phi-enabled applications and libraries on Stampede. The Stampede User Guide has been updated with a list of these applications and libraries, descriptions of which Stampede modules to load, and links to documentation regarding their use.

https://portal.tacc.utexas.edu/user-guides/stampede#software

Computational Chemistry
- LAMMPS: molecular dynamics
  - module name: lammps/09Oct14
  - documentation: http://lammps.sandia.gov and module help lammps/09Oct14
  - execution model: offload; reported 1.8-5x speedups with 12 core host
- NAMD: molecular dynamics
  - module name: namd/2014_07_31-mic
  - documentation: module help namd/2014_07_31-mic
  - execution model: offload; reported 2x speedups
Weather forecasting
- WRF: weather research and forecasting
  - module name: wrf/3.6
  - documentation: http://www2.mmm.ucar.edu/wrf/users and module help wrf/3.6
  - execution model: symmetric mode (host and MIC-side MPI)
Productivity Languages and LIbraries with Automatic Offload to MKL for Dense Linear Algebra
- MATLAB: high-level numerical computing language and environment
  - module name: matlab
  - documentation: https://portal.tacc.utexas.edu/software/MATLAB
  - other comments: ?Bring Your Own License? usage model
- Python: interactive object-oriented language and tools
  - module name: python/2.7.6
  - documentation: https://portal.tacc.utexas.edu/tutorials/automatic-offload
- R: high-level statistical computing language
  - module name: Rstats/3.0.3
  - documentation: https://portal.tacc.utexas.edu/tutorials/automatic-offload
Other Libraries and Tools
- NetCDF3: file i/o
  - module name: netcdf/3.6.3
  - documentation: http://unidata.ucar.edu and module help netcdf/3.6.3
  - execution model: host and MIC-side libraries available
- NetCDF4: file i/o
  - module name: netcdf/4.3.2
  - documentation: http://unidata.ucar.edu and module help netcdf/4.3.2
  - execution model: host and MIC-side libraries available
- HDF5: file i/o
  - module name: hdf5/1.8.13
  - documentation: http://hdfgroup.org and module help hdf5/1.8.13
  - execution model: host and MIC-side libraries available
- NetCDF4: parallel i/o
  - module name: parallel-netcf/4.3.2
  - documentation: http://unidata.ucar.edu and module help parallel-netcf/4.3.2
  - execution model: host and MIC-side libraries available
- HDF5: parallel i/o
  - module name: phdf5/1.8.13
  - documentation: http://hdfgroup.org and module help phdf5/1.8.13
  - execution model: host and MIC-side libraries available
- DDT: debugger
  - module name: ddt
  - documentation: https://portal.tacc.utexas.edu/software/ddt and Allinea and Intel online resources
- VTune: profiling
  - module name: vtune
  - documentation: http://software.intel.com/en-us/intel-vtune-amplifier-xe
- PAPI: profiling
  - module name: papi/5.3.0
  - documentation: module help papi/5.3.0

We anticipate additional Xeon Phi-enabled community codes to become generally available over the next several weeks, including QCD and electronic structure codes. If you are interested in other applications, please note that ?developer? versions are often available several months before the general public release.

Please submit any questions that you may have via the TACC Consulting System. http://portal.tacc.utexas.edu/consulting/

To manage your TACC Announcements subscriptions, please visit https://www.tacc.utexas.edu/news/subscribe