150.214.220.244
Español  English  

Logo Alhambra CSIRC

UGR Alhambra CSIRC
Home Queue Settings and Status

Queue Settings and Status


Alhambra Queues
Queue Maximum CPU Time Maximum Number of CPUs Maximum Memory per CPU Architecture
NOParalela 240 hours 1 1.8 GB Intel 16 cores
12H 12 hours 256 1.8 GB Intel 16 cores
24H 24 hours 256 1.8 GB Intel 16 cores
72H 72 hours 256 1.8 GB Intel 16 cores
24Hbigmem 24 hours 64 7.6 GB Intel 32 cores
72Hbigmem 72 hours 64 7.6 GB Intel 32 cores
120Hbigmem* 120 hours 32 7.6 GB Intel 32 cores
media 2 weeks 160 3.8 GB AMD 16 cores
larga 3 weeks 860 1.8 GB AMD 4 cores

Special Queues
Queue Maximum CPU Time Maximum Number of CPUs Maximum Memory per CPU Architecture
visitante** 4 hours 32 1.8 GB Intel
12Hgpu 12 hours 48 1.8 GB Intel + TESLA
24Hgpu 24 hours 48 1.8 GB Intel + TESLA
72Hgpu 72 hours 48 1.8 GB Intel + TESLA

* Only one job at a time is allowed on queue 120Hbigmem.
** Exclusive use for new users without registered research projects.

Usage Notes

Where are now the UgrGRID resources?

Now, both Alhambra (Intel architecture, SandyBridge 2013) as UgrGRID (AMD architecture, Barcelona 2007) nodes are accessible through a single ssh access point, alhambra.ugr.es. However, these resources have to be accessed through distinct queues. Queues media, larga and muylarga have been set up for this purpose, with maximum job time of one, two and three weeks, respectively.

Important: Queue media gives access to the 16-core nodes that previously were forming the ancient UgrGRID bigmem queue.

Shared-Memory Applications

For shared-memory applications, the maximum number of processors is 4 for queues 12H, 24H and 72H; and 16 for the queues offering a large amount of memory (bigmem ones), such as 24Hbigmem, 72Hbigmem and 120Hbigmem.

On the other hand, on Alhambra, the maximum number of processors is 16, for queues 12H, 24H and 72H, and 32 processors on queues 24Hbigmem, 72Hbigmem y 120Hbigmem.

Distributed-Memory Applications

The maximum number of processors for jobs is indicated in the table above. In addition to this, on UGRGrid, jobs must be submitted with a number of slots multiple of 16 for bigmem queues and 4 for the rest (including the visitante queue). On Alhambra, the number of slots must be multiple of 32 for bigmem queues and 16 for the rest of queues (including the visitante queue).

Non-parallel Applications

To run non-parallel applications can only use the NOParalela queue.

Usage restrictions

Initially, users only have access to resources associated to the visitante queue. In order to be allowed to use the rest of the queues, please fill in the information about the research projects using the supercomputing systems in which you are involved.


Contact us  Site map  Suggestions  Visit us          CSIRC                 © Universidad de Granada