Most effective use of network computing ?

[nazg]

Hi,

I set up a small network computing system, and I was wondering what was the most effective way of using it at its full potential ?

I read the manual, and launched some test, but I am still not quite sure how it works.
Say I have big building interior for which I have around 1000 photo allocated to four arbitrary spatial entities. Those arbitrary spaces are church spaces like : entrance, nave, choir, etc...

The right thing to do would be to make a chunk for every spatial entity, and work them independently throug alignment, dense cloud and meshing, until I bring them back together at the end ?

In fact I quite don't understand how each node is summoned. I thought  (naively perhaps) that I could launch an heavy task related to a big single chunk, and each node would take a part of the computing. Or maybe my setup is wrongly configured, because every time I launch a task, only one node is working while all ther others are desperately idling as "ready".

To be honest, it would be really helpful to add some processing tips into the manual, or some kind of showcase, like the doll or statue.

Best,

[Alexey Pasumansky]

Hello nazg,

The benefit will be significant if you are processing multiple small chunks or frames in the single project, or processing large set of images (thousands). But please note that fine level task distribution is only supported for the Align Photos, Build Dense Cloud and Export Tiles operations. So if you haven't enabled fine level distribution in the Client GUI (Network preferences tab) or building mesh or texture, the processing will involve only a single node.

Also if the task is very small (only a couple of images, for example), the processing could not be split anymore and only one task will be used. You can find the max. possible number of nodes for the current task in the Network Monitor as a second value after slash.