Site Tools


tcr_hardware

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
tcr_hardware [2020/04/20 10:27] tmatejuktcr_hardware [2023/08/01 01:08] (current) – external edit 127.0.0.1
Line 1: Line 1:
 x86_64 nodes : cnodesNNN, gnodesNNN, mnodesNNN x86_64 nodes : cnodesNNN, gnodesNNN, mnodesNNN
-  * CPU: 2x Intel(R) Xeon(R) CPU E5-2680 0 @ 2.70GHz +  * CPU: 2x Intel(R) Xeon(R) CPU E5-2680 0 @ 2.70GHz (all) 
-  * MEM: 64GB<sup>1</sup> (cnodes/gnodes) 512GB (mnodes) +  * MEM: 64GB<sup>1</sup> (cnodes/gnodes) 
-  * GPU: Tesla K20m ( gnodesNNN +  * MEM: 512GB<sup>2</sup> (mnodes) 
-  * NET: ethernet 1Gb, infiniband<sup>2</sup>+  * GPU: Tesla K20m ( gnodes 
 +  * NET: ethernet 1Gb, infiniband<sup>3</sup> (all)
  
  
Line 10: Line 11:
   * MEM: 500 GB   * MEM: 500 GB
   * NET: ethernet 1Gb   * NET: ethernet 1Gb
 +
 +storage :
 +  * shared, about 55 TB ( IBM store Wize family )
 +  * each node has about 800 GB of local hdd stata storage
 +
 +
 +Available constrains :
 +^ constrain name ^ nodes ^
 +| intel | nodes with Intel processor |
 +| gpu | gnodes nodes |
 +| bigmem | mnodes nodes |
 +| ib | nodes with infiniband interface |
 +| power7 | pnode001 |
  
  
Line 15: Line 29:
  
 <sup>1</sup>) on some nodes due to hardware errors available memory is less than 64GB, always use slurm's --mem, or --mem-per-cpu \\ <sup>1</sup>) on some nodes due to hardware errors available memory is less than 64GB, always use slurm's --mem, or --mem-per-cpu \\
-<sup>2</sup>IB is not available on all nodes, use slurm's --gres +<sup>2</sup>) on some nodes due to hardware errors available memory is less than 512GBalways use slurm's --mem, or --mem-per-cpu \\ 
 +<sup>3</sup>) IB (infiniband) is not available on all nodes
  
tcr_hardware.1587371279.txt.gz · Last modified: 2023/08/01 06:38 (external edit)