Software for GPUs inc. compiler/directives, maths libs & tools (debuggers and profilers)
University of Manchester researchers have access to GPU resources at both the University and further afield as listed below. Software applications and programming resources are listed at the bottom of the page.
If you are considering purchasing a GPU card, either for your desktop, workstation or a cluster such as the CSF, please speak to us in advance since we may have access to discounts eg via our NVIDIA CUDA Research Centre status
Resources at University of Manchester
The below information is also available as a Powerpoint presentation
The GPU resources on the CSF, the development cluster Zrek, & RedQueen are available to CSF/RedQueen contributors and via pump priming. The resources on Dali are available to all. For further information please email firstname.lastname@example.org
Shared Facility (CSF): This facility includes 27 NVIDIA GPUs.
- 7x M610x blades, each comprising one NVIDIA card: 2x M2070 and 5x M2050.
16 NVIDIA M2050 with a pair hosted on each of 8 AMD compute nodes interconnected with InfiniBand.
- 2 nodes, each have 2 NVIDIA K20 cards
Full details on the CSF GPU web page
- K40s and K20s
- 2 Xeon Phi cards
- 2 FPGA facilities
RedQueen facility includes
Dali, main visualization resource in IT Services, has a GPU. Depending upon user requirements we use of the following
AMD FirePro V7800 2GB GPU
Nvidia 768MB Quadro FX4800 (temporarily replaced by the above FirePro GPU)
Please see Dali page for latest details
Other Hardware Resources
- 372 NVIDIA Tesla GPUs
- charges apply
- various PRACE partners have GPU clusters available to researchers in Europe (details to follow)
- cloud based
- NB RAC has initiated discussion with various vendors