I take this isn't like the old GPU limit thing and it's an unkown to a lot of people?

If you don't have any automation software you shouldn't run 100 of these.

It's just idiotic to think that you should run 30+ mhash or 100+ devices on one system.

Spread the control out to 10-64 per control system and develop some automation tools or use the ones I post all over for linux.  



chanberg - it's easier than groups of 20 if you label the equipment.


jmordica - Stop being an idiot.

If you are really really planning to run 1000+ of these devices with no automation, no labeling, and generally no clue I hope they catch fire.

Stick 10-64 per control system because that leaves room for a margin of error and for known statistical events of computational error.

LABEL the cables (power/usb) and the gridseeds with the serial number.

Run reports continuously on which devices are up and which are down.

It's just like IBM doesn't make a datacenter and just throw computers into it. They label the rows and cables, and computers.

What's wrong with you.