Has anyone else experience trying to run two different cards in their machine? Any feedback on driver issues/clashes/general hassle would be appreciated.
I thought the Titan X was the 2-in-1 answer, until I read that the double precision wasn’t as the original Titan: http://www.anandtech.com/show/9059/the-nvidia-geforce-gtx-titan-x-review/15
Really need a solution that is good for large architectural models far from origin, and quick GPU renders in Cycles (but without the cost of an m6000).
Anyone using GTX cards for large models in Rhino?
Are quadro’s really better than GeForce on far from origin models? I have not noticed, but then I never work with far from origin models, we always set up a project origin.
I really don’t know, I’ve got no prior experience with GeForce cards - It doesn’t sound like it!
I’ve only used Quadro’s previously, but now I need a more effective rendering option the GeForce are hard to not take a second look at.
I didn’t know it was possible to ‘project an origin’?!
…am I being paranoid about double floating point precision and it’s benefit for Rhino?
No, what we do is that we put a mark at origin with text data information about where that point is in real life, and also a true north arrow if the project is rotated (large buildings with 90 degrees angles are easier to draw when aligned with X and Y)
Many of the architects we work with uses GeForce cards as many softwares now uses DirectX instead of OpenGL, according to them there are no real benefits in spending the extra cash for those apps.
I think the double point precision are mainly for precision calculations (medical and engineering) and not for modelling. Anybody correct me if I am wrong!
Thanks for that tip, makes perfect sense…
I do a lot of work with 3D city models, which can be very far from origin and contain lots of meshes, so have used Quadro to be safe as I thought this would prevent the display becoming scrambled. - Or am I getting confused with double precision meshes here!?
A TitanX card probably would be ideal if the single/double floating point is not an issue.
What card do you currently use Holo?
I am not sure, I wonder this my self!
I currently use a Quadro 4000 on one machine and a GTX 970 on another (and a 330m on my old macbook running windows 10) but I have not compared them against each other on real world coordinates, as I just presumed they all had the same issues)
Thanks for that… Have found researching the gpu options quite a challenge to be honest. Lot’s of differing views/opinions, so helpful to hear from someone who has both.
You don’t find a noticeable benefit with the Quadro when handling lots of objects regardless of origin?
Often the advice on GPU seems to be, ‘depends what you do’. Trouble is, as Rhino is so flexible, I do a bit of everything with it, but I want to compromise as little as possible on performance with this build.
Well, both has their benefits so I’ll do a real life test between the two one of the upcoming days, ok?
(Quadro is faster in shaded mode and handles many and large meshes faster, 970 is faster in rendered mode)
That would be fantastic Holo!!
Look forward to seeing what you find…
I found this article interesting, but was unsure if it meant I could potentially use a Quadro as the main card for modelling, but have a GeForce sitting in there to give Cycles a real boost when rendering?
Thanks for everything so far…
I have been using two GPUs in my computer with Rhino 5 already for two years without problems. I have a Quadro K2200 connected to my screens and a GeForce GTX 780 for rendering. I am using Octane for rendering. These cards have worked well with an old computer, with a Q6600 cpu, as well as on my newer dual Xeon machine.
That’s good to know, thanks for the feedback.
Did you do anything in particular when setting up your system? I did read somewhere about installing the Quadro driver first?
I do not remember exactly how I did with the drivers. I think I uninstalled the Geforce driver when I got the Quadro. Then installed Quadro drivers first. I was then running Windows 8 and 8.1. Now I have updated to Windows 10.
I have been experimenting with GPU rendering. I have both Octane and Arion licenses. I have also tested Blender cycles. Still I prefer Maxwell in my daily work when I render architectural interiors with a lot of spotlights and indirect light. Maxwell runs quite nicely on my 2 x 6 core Xeon.
Thanks for the feedback Markus.
I am thinking of this route because of GPU rendering too. I’ve yet to experiment with the software options but Arion is top of my list, I am typically using Cycles in Rhino at the moment.
Do you see any benefit from having the Quadro as opposed to just relying solely on the GeForce. I too do quite a bit of architectural work, although it is more modelling than rendering.
I do not know if you need a Quadro. However, I think that you need one GPU in your computer which is used for rendering. In this way you can work normally as the compute GPU(s) is crunching numbers. Earlier I was using to Geforces. When I upgraded to a 4K display I got the Quadro.
Octane has now managed to add a feature that lets you set a priority for the rendering on an individual card. In the beginning, trying to work on a single GPU while rendering in the background was a pain and laggy. A dedicated display card was a solution for that. Still is but a dual card setup like where my old GT 460 works great for display and my GTX 780 is very fast for Octane rendering. I think the 460 was the last card that made before the drivers were crippled to make people opt for Quadro. I probably don’t even use those hacked drivers on the 460 though. I’m a big fan of Octane even thought I bought Thea when it came out. Haven’t really used it enough to compare. I think the feature that made me buy it (at the early bird discount!) was the ability to use Rhino clipping planes.