Problem in idc-training-gpu-compute-01 node.

[black__edgar]

Hi all,

I am working in the Intel Tiber Developer Cloud, specifically in a GPU training session, using the idc-training-gpu-compute-01 node.

This node, which has been assigned to me twice in the two days, has a strange behavior. When you run a program (in my case a sycl program)  the program finishes, but the control is not returned to the user; it just freezes.

A quick reproduction of the problem can be seen just by running the "clinfo -l" command.

Has anyone noticed this behavior in the idc-training-gpu-compute-01 node or any other node?

Thanks,

Edgar Black

[Wan_Intel]

Hi black__edgar,

Thanks for reaching out to us.

I've tried reproducing the problem by running the clinfo -l command using the idc-training-gpu-compute-05 terminal. Here is the result when running the command:

 

Please try the suggestion as follows and see if the issue can be resolved:

• Close the existing Terminal and launch a new Terminal in the idc-training-gpu-compute node
• Use a different Web Browser
• Use Incognito mode

Regards,

Wan

[black__edgar]

Hi Wan,

Thank you for answering my question.

The issue is (or was) happening in node idc-training-gpu-compute-01. I was assigned to that node (by chance) twice in the last two days.

My solution to the freezing problem was to log out of the Intel Tiber Developer Cloud and wait for a while before logging in again in the hope of being assigned to a different node.

In the past, I noticed this strange behavior in other nodes. In the last two days, it has been happening in node idc-training-gpu-compute-01.

Thank you again,

Edgar Black

[Wan_Intel]

Hi black__edgar,

Thanks for the information.

Could you please share the screenshot or video of running the clinfo -l command in node idc-training-gpu-compute-01? We'll require this information to investigate this issue further.

Regards,

Wan

[black__edgar]

Hi Wan,

Every time I log into the Intel Tiber Developer Cloud I am assigned a node in a pseudo-random way. I cannot (or I don't know how) access a particular node. I noticed the problem in the idc-training-gpu-compute-01 node because It was assigned to me, not because I chose it.

Therefore, it would be tough for me to produce a screenshot or video of running the clinfo -l command in node idc-training-gpu-compute-01.

The screenshot you want me to produce would look like the one you showed before except that the control prompt never returns to you, and you end up having a terminal in which you can not issue any new command.

Something like:

u2517f30xxxxxxxxxxxxxxx@idc-training-gpu-compute-01:~$ clinfo -l
Platform #0: Intel(R) OpenCL
`-- Device #0: Intel(R) Xeon(R) Platinum 8468V
Platform #1: Intel(R) OpenCL Graphics
+-- Device #0: Intel(R) Data Center GPU Max 1100
+-- Device #1: Intel(R) Data Center GPU Max 1100
+-- Device #2: Intel(R) Data Center GPU Max 1100
+-- Device #3: Intel(R) Data Center GPU Max 1100
+-- Device #4: Intel(R) Data Center GPU Max 1100
+-- Device #5: Intel(R) Data Center GPU Max 1100
+-- Device #6: Intel(R) Data Center GPU Max 1100
`-- Device #7: Intel(R) Data Center GPU Max 1100

I hope this helps.

Edgar Black

 

[Wan_Intel]

Hi black__edgar,

Thanks for the information.

Let me check the issue with the relevant team and we'll provide an update as soon as possible.

Regards,

Wan

[Wan_Intel]

Hi black__edgar,

Thanks for your question.

After checking with the relevant team, we're glad to know that your issue has been resolved. If you need additional information from Intel, please submit a new question as this thread will no longer be monitored.

Regards,

Wan