Unable to access to Iris xe

[ManuelCostanzo]

Hello!

I have a problem accessing to Intel Iris node.

I'm running 

qsub -l nodes=1:iris_xe_max:ppn=2 -d . -I

I get qsub: waiting for job 1938323.v-qsvr-1.aidevcloud to start, but nothing happens. Also I tried with

qsub -l nodes=1:i9-10920x:ppn=2 -d . -I

and the same happened. Accessing to a common GPU worked, but I need an Iris one.

Also, it would be very useful for me to access to a double GPU, but when I do

qsub -l nodes=1:gpu:dual_gpu:ppn=2 -I

I get:

qsub: submit error (Job exceeds queue resource limits MSG=cannot locate feasible nodes (nodes file is empty, all systems are busy, or no nodes have the requested feature))

How can I solve both issues?

Thank you very much!

[Rahila_T_Intel]

Hi,

Thank you for posting in Intel Communities.

Please be informed that sometimes it may show waiting due to unavailability of those nodes. Try after sometime resolves this issue.

We tried to access Intel Iris node using the command you mentioned:

qsub -l nodes=1:iris_xe_max:ppn=2 -d . -I

We were able to get the iris node without any issue. 

You can find out the iris nodes available to you by running the following command:

pbsnodes | grep iris -B 4

Example output: 

s011-n001

   state = job-exclusive

   power_state = Running

   np = 2

   properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,gpu

--

s011-n002

   state = free

   power_state = Running

   np = 2

   properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,gpu

--

s011-n003

   state = free

   power_state = Running

   np = 2

   properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,gpu

--

s011-n004

   state = down

   power_state = Running

   np = 2

   properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,gpu

--

PFA the screenshot of output.

 

Here you can see whether the nodes are free or not.

Then you can request for the available node by using the specific node number.

qsub […] -l nodes=[node_name]:ppn=2

Please refer the below link for reference:

https://devcloud.intel.com/oneapi/documentation/job-submission/

Regarding access to a double GPU:

==================================

We are no longer hosting dual gpu systems on the Public DevCloud. This may change in the future however no ETA at this time.

If this resolves your issue, please accept this as a solution as it might help others with the similar issue. 

Please let us know if your issue still persists.

Thanks

[ManuelCostanzo]

Hi, thank you very much for your answer.

I tried again with that command but again, I am waiting infinitely and nothing happens.

I tried accessing to a specific free Iris node, for example with the command

qsub -I -l nodes=s011-n006:ppn=2

And was the same. So maybe is something related with my account? My ID is u156642.

Thank you again.

[Rahila_T_Intel]

Hi,

We are working on your issue internally and will share you the updates.

Thanks.

[ManuelCostanzo]

Thank you! I'm waiting for your answer

[ManuelCostanzo]

Hi, sorry, do you have any updates? I still can't enter to an Iris node

[Rahila_T_Intel]

Hi,

Thank you for letting us know of this issue, we are working on fixing it by removing older systems with new ones soon. It could take some time, meantime please continue using other GPUs available, thank you for your patience and support.

If this clarifies your query, make sure to accept this as a solution. This would help others with similar issue. 

Thanks

[ManuelCostanzo]

Hey, thank you! Do you know aprox how long will take this update? Can you let me know when is fixed? Thanks

[Rahila_T_Intel]

Hi,

The current DG1/Iris_Xe_max nodes are very unstable. We are in the process of getting a equivalent replacement that will support the Gfx needs. ETA 08/15/2022.

Please let me know if we can go ahead and close this case? 

Thanks

[ManuelCostanzo]

Perfect,  thank you! Yes close, I'll wait for the changes.

[Rahila_T_Intel]

Hi,

Thanks for your confirmation.

If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Thanks