Tesla GPU computer
The Tesla S1070 GPU based computer is a 960 core, highly parallel programming environment, using 4 NVIDIA 1060 Graphics cards. The attached Linux computer has the basic tools required to programme the NVIDIA cards to process highly parallelised programs.
The tools include device driver extensions to handle the cards, as well as CUDA programming extensions to PGI FORTRAN and C. Apart from the PGI programming tools you may find other CUDA enabled applications useful to you. If you do, please let us know and we will be happy to install them on the system.
This environment is offered as a sandbox for researchers to learn and practice GPU programming. As such, applications are limited but you can use the server for any programming tasks that you may have.
Access
If you wish to have an account on the Linux/GPU system please send an email to: eresearch@latrobe.edu.au
Further information
Find out more about NVIDIA GPU programming.
Find out more about Tesla computing.
The system comprises
One server controls 1 part of the GPU cards so there are, in fact, 2 servers you can use.
| Name | Running applications | Cores | Memory |
|---|---|---|---|
| Tesla S1070 | GPU programmes | 240 core (960 cores in total) | 4MB Ram per card (with 4 Graphics cards) |
