I’m on Windows, and Dr. Queue isn’t supported for that platform yet. I’ve been using Piovra, a python-based renderfarm. It’s working pretty well, but multi-processor support isn’t quite working under Windows yet.
The render daemon button doesn’t do anything at the moment. These are programs and/or scripts that can take and distribute blender files over multiple computers to increase rendering speed of animations. You can integrate Piovra right into the text window of Blender. I’m not sure about Dr. Queue, but it looks like that is a separate program.
so i have to have multiple cpu or pc!? ahh no thanks - to expensive even for me! someday when my studio gets bigger i plan on getting one of the multi processor amds which will really make rendering faster!
I just looked at Nitrox. Seems good and stable, but it doesn’t have the best Server-Client connection scheme. The master should ALWAYS have the list of the clients, not the other way around. In Nitrox, the clients are constantly pinging the server to see if there’s a new scene. This not only creates excess network traffic, but only allows you to have a single master server. I still prefer Piovra because you put all the client’s IP addresses in the master and it contacts them only when there’s a new scene. And also this way you can have two or more computers act as servers and utilize the same render farm.
I am not sure if Dr.Queue can have multiple Masters but it sounds more or less the same as Piovra.
I wasn’t shocked by Nitrox either, but in its defence I really have not used it a whole lot.
Dr. Queue has worked for me and I am trying to help the developer as I can…
I am still shocked that more people have not voted/discussed what they are doing. I suspect that either more people are using farms and just not piping up, or they are not taking advantage of the systems out there.
If there are others I forgot to list, bring 'em up.
As I think about it, though, I’m not sure how well that would work for me if I had very large textures. I am sitting on one right now that is slightly better than a gig(A long animation we are using inside another animation.) And shuffling that all over the net might not be the best idea. If you had many machines to push the data to, and you are using a small pipe (read average dsl or cable) you might spend longer shuffling the data to the clients than rendering!
In this case I see an advantage for remote mounting the data (like I do with nfs and Dr. Queue) Is that possible with Xgrid or does it copy all the data at once? I would like very much to see it in action.
I’d see integrating a Bit Torrent client into a renderfarm setup, so the master computer would send the information to 2 computers at a time, and those computers would tell 4, and those 4 would pass that on to 8, etc…
Definitely. Xgrid shouldn’t be used if you are only rendering low resolution, short length animations. It will take just as long to transfer the data, even over ethernet, as it would to just render it on one machine. For example, don’t think rendering an animation at 160 x 120 that is 10 frames long will do anything.
On the other hand, if you are rendering an animation that is 10 seconds long (300 frames for video) Xgrid will greatly speed up your work.
Currently (remember, Xgrid is still beta) Xgrid distributes all of the information to the agents when the project starts, or if new agents join the grid they are sent the data to begin work. After that the agents just crank it out.
So far I’m pretty happy with it because I can also use it with computers that are otherwise dedicated to other use. When my father leaves work at night and the screensaver comes on, his computer joins my grid. That’s about 12-14 hours of computing power I get to use that would otherwise be wasted.
Xgrid now has agent software for UNIX machines to work as agents, although the software is 3rd party. But it works, and I suspect that software will soon show up for Windows machines to work as agents as well.
I love the fact that anyone can lend their computer to my grid if I give them my IP address. It’s a great way to grab computing power from your family and friends