The DevCloud documentation mentions that the nodes are connected to shared NFS storage:
"... In the Intel DevCloud for oneAPI Projects, compute nodes are interconnected with a Gigabit Ethernet network; connection to the NFS-shared storage has higher-speed uplinks. ..." (Queue Management | Intel® DevCloud)
Are any technical details available on the NFS-shared storage? Is it running Lustre? Are there any user-configurable settings or usage patterns that affect performance?
In addition, is there any storage local to the nodes? Is $PBS_SCRATCHDIR (/local) physically attached to a compute node, or is it on the same NFS-shared storage hardware as /home?
How should $PBS_SCRATCHDIR be used? Should all data be processed by a node first be copied to $PBS_SCRATCHDIR?
Thanks for posting in Intel forum.
We have answers for few of your questions. For the remaining questions, we are checking internally to get more information.
For your question regarding "user-configurable settings", it is not user configurable. Regarding your local storage question, we recommend you to use the local storage on each node.
Hi @User01 ,
We got an update from the admin team regarding this issue. All the documentation related to home directory and local storage is here: https://devcloud.intel.com/oneapi/documentation/file-storage/
Please go through the above documentation and let us know if we can discontinue monitoring this thread.
If this resolves your issue, make sure to accept this as a solution. This might help others with similar issue.
We assume that your issue is resolved. If you need any further assistance, please post a new question as this thread will no longer be monitored by Intel.