User Tools

Site Tools


bayes:cluster

Differences

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

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
bayes:cluster [2014/09/04 14:55]
wphase [GPU nodes - Mésocentre de l'Université de Strasbourg]
bayes:cluster [2017/09/14 11:11] (current)
wphase
Line 1: Line 1:
-====== ​Project BAYES - Resources ​======+====== ​Computational resources ​======
  
 ==== GPU nodes - Mésocentre de l'​Université de Strasbourg ==== ==== GPU nodes - Mésocentre de l'​Université de Strasbourg ====
  
 == Current configuration == == Current configuration ==
-  * publicgpu or grantgpu queue:+  * NVIDIA K80: 
 +    * funded by the IDEX project BAYES 
 +    * 5 compute nodes with 4 K80/node - 40 GPUs 
 +    * nodes hpc-n443-446,​ hpc-n517 
 +    * pri2015gpu partition 
 +  * NVIDIA K40: 
 +    * funded by the IDEX project BAYES + NVIDIA donation 
 +    * 2 nodes with 4 K40 GPU/node - 8 GPUs 
 +    * nodes hpc-447-448  
 +    * pri2015gpu partition 
 +  * NVIDIA K20:
     * 2 nodes with 4 K20 GPU cards (2 from Z. Duputel INSU project)     * 2 nodes with 4 K20 GPU cards (2 from Z. Duputel INSU project)
-    * 1 nodes with 2 K20 and one K40 GPU card (NVIDIA donation) +    * publicgpu or grantgpu queue 
-  * public queue:+  * Other (public queue):
     * M2050, M2070 and K20 GPUs.     * M2050, M2070 and K20 GPUs.
- 
-== Planned configuration (28 GPUs) == 
-  * Must be compatible with existing NEC nodes of the HPC cluster 
-  * The GPUs must be able to host G and M matrices larger than 5GB. 
-  * 6 compute nodes 
-    * 4 GPU NVIDIA K40 per node (total: 24 GPUs) 
-    * Each GPU will have its own dedicated PCI-E slot 
-  * 4 GPU cards to be installed on existing nodes (2 GPU/node) 
-    * NVIDIA K40 
-    * Each GPU card will have its own dedicated PCI-E slot 
-  * We also request 3 years of software and hardware support 
-  * Budget: 147 000 euros VAT included 
-  * Currently working with Transtech, Nec and Dell for pre-pricings : 
-    * Nec/​Transtech working on the same nodes as the ones installed 
-    * Dell working on a solution based on a C8000 chassis and C8220x nodes. If the C8000 chassis is full, could be the same density as the NEC nodes. Otherwise may be less interresting 
bayes/cluster.1409835357.txt.gz · Last modified: 2014/09/04 14:55 by wphase