OTOY Forums

I'm going to build a new system and think of buying one of the new nVidia geforce gtx 560 ti graphic cards for using blender and octane render. My system specs will be:

Asus P8P67 pro
Intel i5-2500K
DDR3RAM 3x 2GB DDR3-1600 Corsair XMS3
Samsung SpinPoint F4 Desktop Class HD322GJ 320GB
Philips Brilliance 225P1 monitor (1680x1050)

Would the 560 ti be the right thing to go or should I better buy a quadro 2000 professional card? Any experiences? The 560 ti is available as 1GB and 2GB version, while the 2GB from Gainward has a phantom cooler and uses 2,5 slots, which is possibly not the right thing for an sli-connection in the near future. I read there's also the possibility to use one gtx and one quadro together. What would you suggest?

Hi,

The best thing to do is to buy both and compare them because , I believe, no one here knows .
Plus once you compared them could you post the results .. ?

Sorry just being cheeky lol, I personally went for the 560 TI because it is better for octane however worse for 3d aps such as 3ds max but since Im going to be learning I guess that I wont mind the bottleneck.

the quadro 2000 has 192 cuda cores vs the 560 which has 384, this means that for GPU computing the 560 is twice as fast, so if Octane is your main focus, go for the 560, now if your main focus is 3d modeling quadro surpases by far the gforce in most apps, if you have the money you could go for both, the quadro for working and display and the 560 as dedicated render GPU, if you don't I would still buy a cheap GTX 2xx series for display, you'll experience a much more ease at rendering and working at the same time. Heck I can play black ops while rendering... I've said it before, but it's so awesome I can hardly believe it

Many thanks for the fast response! I think I'm going for a 560 ti now and a quadro later, because I ain't got the money for both at the same time. If it's only about cuda cores for rendering I just discovered the geforce gtx 470 got 448 , so anyone got experiences with this card, maybe even in comparison to the 560 ti? What about the Gainward gtx 560 ti phantom 2GB?

And what about the gtx 285 (sorry, maybe I'm driving it to far) ? It has 240 Cuda Cores and "Kubo" mentioned the gtx 2xx series works better for display and rendering than the newer gtx cards. It's 2 years old and quite expensive tho.

I didn't mean that the 2xx series work better than the 4xx or 5xx, just that they are cheaper, and since you don't want it for rendering just display you won't need a "great" gpu, for rendering you buy a second gpu and have it dedicated and go for a 460 or 560 with 2gb if you can afford it. The extra vram will be put at good use in octane.

kubo wrote:I didn't mean that the 2xx series work better than the 4xx or 5xx, just that they are cheaper, and since you don't want it for rendering just display you won't need a "great" gpu, for rendering you buy a second gpu and have it dedicated and go for a 460 or 560 with 2gb if you can afford it. The extra vram will be put at good use in octane.

Very good advice, I´m useing a HD5870 as my second card since it´s draw less power(only 850w powersupply)

But is 12gb ram enough?, I´m on 8gb, and upgrading to 16gb (fear it´s to limited).
After some testing with huge terrains with many trees/cars/houses/etc! I run out of mem everytime.

GPU mem is only at 500-700, I then add some more geometry and it´s out of mem with CPU?

Think I´m gonna build a new PC to test, but if any of you know a limit to a single wavefront obj file I would like to know
before I waste money on ram/CPU.

No problem waiting for a 2gb file to import, as long as it´s not crashing.

As you can see in my sig I have 8gb ram and this is enough to fill 2gb vram entirely with geometry however I think it pages about 1gb to disk (not too sure about interpreting Win7 Task Manager stats yet vs XP I am used to) - and it does take a while to load.
If you have a newer system with triple mem slots 12gb ram should be good for 2,2.5,3gb vram in practical use. 16gb would be more suitable for 4gb,6gb I think, again not sure someone needs to try maxing out these cards now we have Cuda3.2 and their mem fixes.
Incidentally 2gb vram holds about 10-11m tris geometry only.

But is it ok in 1 mega file that contains all this generated info?

I´m just curious of the endless power if Octane is able to link files like Revit.
Perhaps I´m to focused on xref/proxy stuff, and Octane is able to safe import 5gb of obj files, but is it a smart path with large scale customers in a multi worker enviorment?

Shit!, spelling errors is a bad sign on a Friday.