News & Events
14th,May,2012 : Changes to the SGE Job Scheduler
On 14th of May, we will be introducing improvements to the SGE job scheduler so that we -
(a) get better overall performance from the cluster,
(b) make life easier for users by relaxing some of the currently imposed limits.
After the changes:
- default memory allocation limit for each job will go up from the current 2 GBytes to 6 GBytes.
- Anyone needing larger than 6GBytes of memory will be able to request it via the optional -l mem= parameter as usual.
E.g. -l mem=64G - The stack-size limits will be removed to avoid the risk of excessive recursion causing crashes.
- The risk of exceeding the filestore allocations will be reduced by not allowing the creation of core-dump files by default.
- There will be a new optional -l rmem= parameter introduced to fine-tune the memory usage of large memory jobs.
- At the same time our documentation will be updated to reflect these changes.
10,May 2012: Workshop for Introducing High Performance Computing Using Iceberg
PLEASE NOTE THE CHANGE OF DATE FOR THIS EVENT
This 'mostly morning' workshop is jointly given by software vendors Intel and ALCES Software and will provide the following sessions:
- An introduction to high performance computing
- An overview of the iceberg HPC facility
- Data storage and management
- Running applications and using the scheduler
- The intel cluster toolkit
Anyone wishing to attend this workshop should apply via this link.