Large Memory Nodes

The IBEX cluster has varying large memory nodes from 498 GB to 3 TB from different system architectures. However, part of the memory is used by operating system and remaining memory is usable for the job execution. The summary of large memory nodes are listed in the below table:

 

                                              Table. Ibex Large Memory Nodes

System architecture

Cores per node

Number of nodes

Recommended max memory per node$

AMD Abu Dhabi

64

 1 498 GB

3

750.86 GB

8

988.12 GB

1

1.3 TB

3

1.47 TB

Intel Westmere

64

1

1.84 TB

1

1.9 TB

Intel Westmere

80

1

1.9 TB

Intel Skylake

32

4

2.93 TB

$ The usable node memory represents the available memory for job execution.

The jobs with the larger memory requirement can be submitted using  `--mem=###G` constraint in SLURM job scripts. For jobs to run on a large memory node they must request at least 370,000MB of memory. For any job that requests less than 370,000MB it’ll be classified as a normal compute job and will run on a normal compute node.

For example:

  • When ` --mem=400GB` is specified, the SLURM will select any one of the node which has 400 GB or more memory. I.e., Any one of AMD Abu Dhabi, Intel Westmere or Intel Skylake node will be selected by the SLURM scheduler.

  • When ` --mem=2000GB` is specified, the SLURM will select any one of the node which has 2000 GB or more memory. I.e., One of the Intel Skylake node will be selected by the SLURM scheduler.

 

For further info send us a query using the Contact Us page.

Alternatively, send an email:

  1. Application installation/failure/support: 
  2. System issues/failure/support: