Just a message to inform you that on sirocco[03-04] nodes, only 3 K40M GPU cards are now available instead of 4 initially because of a hardware problem on these nodes.
We are sorry for this inconvenience.
Just a message to inform you that on sirocco[03-04] nodes, only 3 K40M GPU cards are now available instead of 4 initially because of a hardware problem on these nodes.
We are sorry for this inconvenience.
New compute nodes are now available (see https://www.plafrim.fr/hardware-documentation/):
To reserve them, use the constraints “bora”, “zonda” or “rtx8000”.
These machines have just been integrated, do not hesitate to contact us if you see improvements to be made or need further information.
The FAQ has been updated to describe all storage spaces available on PlaFRIM and the quota command to display their usage and limits.
The shutdown is now over and PlaFRIM is available again.
In order to proceed with the annual total electric shutdown of the building scheduled for 25 April, we have to shut down the PlaFRIM infrastructure from 24 April during the afternoon.
A normal situation is scheduled on Monday 27 April during the morning.
Thank you for your understanding.
A new parallel BeeGFS file system (see https://www.beegfs.io/content/) is now available on PlaFRIM. You will find a /beegfs/<LOGIN> directory on all PlaFRIM nodes.
This parallel file system replaces the /lustre storage which is no longer under warranty and is not recommended.
The data in this directory are not saved. The quota is 1 TB per user.
If you need to write/read massively and/or in parallel, this is the preferred file system unlike your /home/<LOGIN> directory.
Information on all available storage spaces can be found in the plafrim website’s FAQ (see here).
Please do not hesitate to contact us if you need further information
New AMD compute nodes are now available (see Hardware Documentation/) :
– diablo [01-04]: 2 * AMD EPYC 7452 32-Core Processor and 256 GB RAM
– diablo05 : 2 * AMD EPYC 7702 64-Core Processor and 1 TB RAM
To reserve them, use the “amd” or “diablo” constraints.
Please do not hesitate to contact us if you need further information.
The network switch was replaced on January 15. Since then, we have not observed any outages.
We confirm that the problems of access to the bora [001-040] and devel [01-02] nodes are solved.
Please do not hesitate to contact us if you need further information.
Many of you are reporting connection problems related to changing the fingerprint of development servers when connecting to SSH.
To make sure you can accept this change, you will find below the new fingerprint of these servers :
SHA256:HTzNYIkxeVcDVORkoKXleJTOEqbtq5gs9UfqjyHBOGY
Do not hesitate if you need more information.
The server room of the Talence building will be undergoing refurbishment work on the cooling system and will require a total shutdown of PlaFRIM from 1st August starting at 12.00 with a restart during the 6th of August.
Thank you for your understanding.