Research Infrastructure

Archive for the 'GPU' Category

N8-CIR GPU Cluster (bede) now open

Wednesday, December 2nd, 2020

Bede, The N8’s new high performance GPU computing platform, is now open for users. Bede is the N8’s IBM Power9 + Nvidia GPU-based high-performance computing (HPC) platform. Unlike traditional x86 platforms, Bede is GPU-based and makes use of NVIDIA’s NV Link high-bandwidth interconnects to move tensor outputs between the GPU and system memory. This unique architecture is ideally suited to artificial intelligence (AI) and machine learning (ML) applications. Bede will also be able to work.. Read more »

RI Update, Jan 2015

Friday, January 16th, 2015

The Research Infrastructure Team If you have any questions about the items below, please contact the RI team at its-ri-team@manchester.ac.uk CSF Refresh The Computational Shared Facility (CSF) is the University flagship computational resource. It started life four years ago and is now running 1000s of jobs for over 650 users from 35 research groups representing all faculties. The CSF is getting a New Year refresh. This will secure the service for years to come. The.. Read more »

iCSF Graphics/Viz Capability

Monday, April 7th, 2014

The iCSF now has a graphics/viz. capability. This capability is based on three AMD GPUs and works with a new Research Virtual Desktop Service (RVMS) node which is tied closely to iCSF nodes: the network latency between this RVMS node and the GPU host is low; and bandwidth is high and uncontended. Furthermore, network traffic from RVMS node to users’ desktop is heavily compressed (the RVMS uses NX/X2GO). The combination of these features means that.. Read more »

CUDA 5.5 available on CSF

Tuesday, September 3rd, 2013

The CUDA installation on CSF has now been upgraded to version 5.5.22 (previous versions are still available). All GPU nodes have had the Nvidia driver upgraded to version 319.37 to support CUDA 5. This version of CUDA includes the Nvidia Visual Profiler (nvvp) as well as the usual compiler tools, numerical libraries and sample codes. CUDA 5 is also supported by the PGI 13.6 compiler available on the CSF. Please see the CSF GPU documentation.. Read more »