I want to use GPU Max Iris on devcloud but I have 2 problems:
1. When I can enter using: qsub -I -l nodes=1:quad_gpu:ppn=2, or qsub -I -l nodes=1:dual_gpu:ppn=2, or qsub -I -l nodes=1:iris_xe_max:ppn=2 -d ., I execute a sample DPC++ program and nothing happen. It's like the GPU cant start or something like that and it freeze and never ends. Also if I launch a job without the interactive mode.
2. Actually when I want to enter to Iris Max, I get: qsub: cannot connect to server v-qsvr-1.aidevcloud (errno=111) Connection refused
If I try using another GPU, all works great (usnig the same program, etc), the problem is with Iris.
Anyone knows what is happening ? Thank you so much !
Your issue might be because the node with your requested feature might not be free at the time of request. To list down the nodes with a particular feature use the below command.
pbsnodes | grep "properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,quad_gpu" -A 4 -B 4
Note: Above command is to list the number of quad_gpu nodes.
You can request node again. Incase if you have any issues in accessing those nodes then search for nodes with the above command. Even after this if your issue persists then let us know.
Thanks for the update.
Kindly provide the command line and screenshot to reproduce your issue. Also we suggest you to list free nodes with pbsnodes | grep "properties = core,cfl,i9-10920x,ram32gb,net1gbe,gpu,iris_xe_max,quad_gpu" -A 4 -B 4 command and login with node number instead of node property.
We 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.