Skip to content

GPU computing

GPU job

To run GPU calculation, the user needs to specify number of GPU cards only. The PBS scheduler will route the job automatically into one of the gpu* queues.

name of GPU queue does not need to be specified anymore

Until the upgrade to Open PBS server the name of the queue had to be explicitly specified to run GPU job. This is not required anymore. Any job with non-zero ngpus parameter will be routed into gpu queue by the scheduler.

available GPU queues Walltime range 00:00:00 - 24:00:00 24:00:01 - 336:00:00

User group iti has a reserved GPU queue

Members of the iti group (Institute of Theoretical Informatics, University of West Bohemia) have their own GPU cluster konos with priority access through direct submit to queue.

PBS resources

gpu mem

PBS parameter gpu_mem specifies minimum amount of memory that the GPU card will have.

qsub -q gpu -l select=1:ncpus=1:ngpus=1:gpu_mem=10gb ...


PBS parameter gpu_cap is Cuda compute capability as defined on this page.


With the introduction of new PBS server running on OpenPBS, the specification of gpu_cap parameter is done in two distinct ways depending whether you submit to PBS Pro or OpenPBS scheduler.


The user can specify a minimal required architecture (compute_XY), or a minimal required version within a given architecture (sm_XY).

Minimal architecture:

gpu_cap=compute_70   # will give you 7.0, 7.1, ... 7.5, but also 8.0, 9.0 ...

Minimal version of a chosen architecture, e.g. 7 ("Volta"):

gpu_cap=sm_72        # will give you 7.2 till 7.5, but not 8.0 and higher

The requirements can be combined in a comma-separated string.


The commas are evaluated as an OR operand.


qsub -l select=1:ngpus=1:gpu_cap=\"sm_65,compute_70\":mem=4gb   # 6.5 or 7.0 and higher
qsub -l 'select=1:ngpus=1:gpu_cap="sm_65,compute_70":mem=4gb'   # dtto


Note that the quotes enclosing the gpu_cap options must be protected against shell expansion either by escaping them or by enclosing the whole qsub command into single quotes.


PBS parameter cuda_version is version of CUDA installed.

System variables

IDs of GPU cards are stored in CUDA_VISIBLE_DEVICES variable.

These IDs are mapped to CUDA tools virtual IDs. Though if CUDA_VISIBLE_DEVICES contains value 2, 3 then CUDA tools will report IDs 0, 1.