Workers are visible but not usable

Install, Configure and Update
1

Dear all,

That relates to my previous post “GPU not available”

I just installed 1 master and 2 workers (several CPU’s and 1GPU each). The are all visible as 3 targets in Resource Manager/Instance Information and since they are all identical I copy here the target 3:

Target 3: gr3 node

Cores 16
Memory 64 GB
GPUs 1
Worker bin path /usr/local/cryoSPARC/cryosparc_worker/bin/cryosparcw
Hostname gri13
Name gr13
Cache path /data-em/cryos/temp
Cache quota (MB)
SSH String cryos@gr13
Cache Reserve (MB )1 0000

I went through the whole tutorial and in no instance I could use any worker besides the one on the master. I have a concept problem here. I was told that majority of programs are using GPU’s only on one machine (by itself limiting) I thought that at least I can choose the machine to work with but I could not. So what is the point of not usable workers? Something is wrong with my setup or with my logic of using workers. Can someone explain the concept?

Best regards,

Ketiw

2

Dear All,
Investigating further I realized that:

  1. All programs are designed to run only on one machine (with multiple CPU’s/GPU’s) at an instance.
  2. To be able to select GPU’s for particular run you need to restart the cryosparcm on the master after connecting workers, which is not mentioned in the guide.
    Hope it will be helpful for newcomers,