Any chance Photoview might get some processor sharing capabilities, like Maxwell or Studiomax?
That's my wish too. I think this isn't a priority for SW guys.
I dont know if it exists as an Enhancement request
I've requested it
The underlying technology is there as modo has the ability. We just have to wait for SW to make it available in PV360.
I just ask my Engineers around me to keep their boxes open so I can remove their processor when I need to do some renderings . Seriously though, this is a good idea not only for renderings but for simulations too.
If i remember correct, someone here said that modo's render farm controller isnt good at all. It cannot use all the processors available at full power. Have you tested modo to a render farm yourself?
Realistic feedback would be highy appreciated.
Not sure about that. I have used the network renderer several times with modo and haven't seen what you are referring to. It's pretty seamless.
Maybe i was wrong. Since you are using it, i am sure its ok.
I must find some time for modo as well.Seams quite appealing.
From my multiple tests with Modo on both PC and Mac systems the "Slave" only completes about 2-8% of the rendering in the same time as the master computer completes the remaining 98-92%. This is also what Modo users say on the forums that I have seen. Even when multiple "slaves" are added people have only posted (Modo Forums) results of an increase of only 20% speed. If others know how to get the slaves to use >50% of the CPU or better than 20% increase in speed over the master computer please pass it on.
I've used the network render option in modo as well with decent results. It's not a 100% speed increase but I believe that has a lot to do with the processing time for each frame you're rendering. The slave system(s) have to load the information for the scenes each time to process their portion, you aren't just tapping into their processing power. Loading this information takes time so if your render takes 5-10 minutes to process normally then the speed increase you'll see from network rendering will be very small. If your scene takes 45-60 minutes to render then the speed increase you see from network rendering will be greater. Of course the other real benefit is rendered animation. Shaving even just a few minutes of tme off per frame can add up to a huge time savings when you're processing 100,s or thousands or frames.
How is "decent results" defined?
To me decent results would be at least 25% of a bump in performance for each slave. 2-5% just does not cut it.
I wish it was faster as I purchased Modo for this purpose only to find this out after.
If others know better how to configure this I would like to know.
Jonathan and Paul,
That is what i was remembering as not taking full advantage of the processing power of each render node.
I see that Rob is telling something similar as well.
It seems that the chance to fully exploit slave cores is a combination of different factors.
I also think that the speed of your master machine has a lot to do with it. In other words, if it's real fast, the slaves may not have time to obtain the info over the network and render them out.
I agree there are several factors contributing to the success but I have seen it work pretty well. Keep in mind that I only use it for animations. Also - I the irradiance caching calculations are performed by the host machines. This happens before each render in an animation - unless you are in walkthrough mode...
It's not perfect and from what I read it's not as robust as other software's version of network rendering but it does work to a certian degree. I can't really give percentages because I just set it and go. I really don't time out the saving. What I personally have done is purchase a 12 core / 24 thread overclocked machine from Boxx. It's blazingly fast and really its a mini render farm all on it's own. In that case you are getting 100% from everythread. It depends on a number of factors but just for general image creation at 2400 pixels or so I'm spitting out stuff in 5 minutes or less. Your typical SW splash screen image (they are small images) will finish processing in 1-2 minutes.
Have a look at the Luxology forums. You be able to see what everyone is saying both good and bad.
Sounds interesting. How does SW and PW360 rendering and animations work on the Boxx?
Great. There is no speed increase for SW as far as modeling but renderings and rendered animations are as I described above. Simulation solving is also faster. The systems aren't cheap but if you're doing a lot of image creation or animation work and waiting 1-2 hours or longer for your render to complete then it doesn't take long for a system like this pay for itself in time savings.
Does not look to good....
I have been using Deadline to manage my modo render farm. Works well managing my render queue.
I have 6 Core i7 2600k systems and a dual socket Xeon X5690 system. The Xeon system uses the same motherboard that is in the Boxx computers. It is basically a Boxx system without the fancy case.
The Xeon machine is setup to render SolidWorks 2011 PV360 files as well as Modo. It can use the 12 cores with 24 threads. About as peppy as you can get with SolidWorks rendering.
I render Modo frames on all the computers, using Deadline to manage the render queue.
I found Modo's solution for network rendering to be pretty slow. It is not efficient at handling all the network traffic required to keep the packets in good order on all the render nodes. The render queue managers such as Deadline or Butterfly do a much better job with Modo.
P.S. If anyone needs to off load their rendering tasks my farm is available for rendering services. www.renderbay.com
I've used Anna's render farm. It was an awesome experience. She was quick and low cost - and of course she just rocks anyhow.
Using the render farm helps me to stay focused on buildling animations by not tying up my machines.
I have outsourced rendering before but Anna's got it setup right. I'd recommend it highly.
Retrieving data ...