GPU:No device of requested type available.

[HYB]

When a queue is defined in a file and bound to a device, but only this device is the GPU selected by the gpu-selector, an error will occur, indicating that there are no available devices.

After I entered the following command, I fell into a waiting loop and did not run correctly.

qsub -I -l nodes=s001-n233:gpu:ppn=2 -d . build.sh

qsub -I -l nodes=s001-n233:gpu:ppn=2 -d . run.sh

 A few months ago can run contains GPU_ The selector file can't run now. The errors are as follows:

Caught this SYCL exception: No device of requested type available.
Please check https://software.intel.com/content/www/us/en/develop/articles/intel-oneapi-dpcpp-system-requirements.html -1 (CL_DEVICE_NOT_FOUND)

[JaideepK_Intel]

Hi,

Thank you for posting in Intel forums.

The nodes which you mentioned in above may or may not available always so it is better to access nodes in a generic way i.e. you can try below commands:

qsub -l nodes=1:gpu:ppn=2 -d . build.sh
qsub -l nodes=1:gpu:ppn=2 -d . run.sh

Thanks,

Jaideep

[JaideepK_Intel]

Hi,

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

Thank you,

Jaideep

[GRN2]

The workaround how to use interactive sessions.

1. Connect to login node using terminal.

2.  Prepare job sh (job.sh) script on your home directory on DevCloud with this content:

#!/bin/bash

sleep 3600

3. Submit the job "qsub -l nodes=1:gpu:ppn=2 job.sh"

4. type "qstat -an" in this terminal and get the your compute node (the last string from output, like "s011-nxxx")

5. Open new terminal on your host machine.

6. Type "ssh s011-nxxx.aidevcloud"

7. Now you got interactive session in the second terminal. Check your sample and target device must be available.

8. You can also create VSCode session using Remote-SSH.

[HYB]

Thank you very much for your reply. However, two tasks cannot be deleted, and the new job.sh cannot be executed.The second terminal connection failed.The error message is as follows:

kex_exchange_identification: Connection closed by remote host Connection closed by UNKNOWN port 65535

[GRN2]

Yes,

this workaround works only for one PBS job on DevCloud.

[HYB]

However, my terminal is currently occupied by tasks that cannot be deleted and cannot be used.Can you help me delete these tasks?Thanks

[GRN2]

qdel all

and then check via qstat

[HYB]

This command cannot delete the task. The error message is as follows

[GRN2]

Check via qstat. The jobs must be deleted.

[GRN2]

Do not pay attention on qdel message.

qdel all works fine. 

[GRN2]

Wow, Elap time is 50-80h.

You can try to connect to compute nodes via ssh (from login node) and check your processes

ssh s001-n169.aidevcloud

[JaideepK_Intel]

Hi,

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

Thank you,

Jaideep

[JaideepK_Intel]

Hi,

I assume that your issue is resolved. If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Thanks,

Jaideep