Ocelote has 46 new compute nodes with Nvidia P100 GPU's. These are available to researchers on campus. There will be fairshare limitations but the intention is for them to be as widely available as possible. There are still compute nodes on El Gato with 70 nodes provisioned with Nvidia Tesla K20's.
Cuda Modules
Currently the following Cuda modules are available on Ocelote:
/cm/shared/modulefiles
cuda75/blas/7.5.18
cuda75/nsight/7.5.18
cuda80/blas/8.0.61
cuda80/nsight/8.0.61
cuda75/fft/7.5.18
cuda75/profiler/7.5.18
cuda80/fft/8.0.61
cuda80/profiler/8.0.61
cuda75/gdk/352.79
cuda75/toolkit/7.5.18
cuda80/gdk/352.79
cuda80/toolkit/8.0.61
/cm/shared/uamodulefiles
cuda75/neuralnet/5/5.1
cuda75/neuralnet/6/6.0
cuda80/neuralnet/5/5.1
cuda80/neuralnet/6/6.0
Applications
Many applications have been optimized to run faster on GPU's. These include:
NAMD
VASP
GROMACS
LAMMPS
GAUSSIAN
MATLAB
AMBER
ANSYS Fluent
ML and DL frameworks
Training
We host workshops from the Pittsburgh Supercomputer Center which is a NSF funded location. We are working with Nvidia to offer a workshop in the April 2018 timeframe.