Due to the availability of newer and more efficient machines, we will stop accepting new grant applications for individuals with resources requested on Prometheus. New grants should apply only for resources on Ares and Athena.
Here are some tips regarding migration from Prometheus to Ares and Athena:
- Ares is a general CPU cluster with a small number of V100 GPGPU cards. Athena is a specialized GPGPU cluster with a significant number of A100 GPGPUs. Information about new machines can be found here: Ares, Athena.
- Please note that Ares and Athena don't use normalization. One hour spend on CPU equals to one cpu-hour on your grant. This change should be reflected in the grant application, where resource numbers on Ares and Athena might be smaller than on Prometheus.
- Ares and Athena are using shared storage, so please apply for storage from "STORAGE-01" site in the PLGrid portal. This storage is available on both machines and allows for easy data sharing.
- Please mind that $SCRATCH and group directories on Prometheus are successively cleaned. If you need to keep your data, please transfer it to Ares or any other storage offered by Cyfronet.
- If you need to transfer files from Prometheus to Ares or Athena, please apply for the space at the destination machine. After obtaining the space allocation, please transfer the data. The transfer can be done simply by using the "scp" or "rsync" commands on a login node, described in more detail here:
https://docs.cyfronet.pl/pages/viewpage.action?pageId=4260613#Sposobydost%C4%99puiprzesy%C5%82aniadanych-2.Kopiowanieplik%C3%B3w
https://download.samba.org/pub/rsync/rsync.1
If your dataset is large, you can consider performing a parallel file transfer:
https://www.gnu.org/software/parallel/parallel_examples.html?highlight=rsync#example-parallelizing-rsync - If you need any help with the transfer, or your data transfer poses a challenge because of the data size or number of files, please get in touch with the helpdesk: https://helpdesk.plgrid.pl