Does Quadro even make sense? I’ve worked with regular gaming cards (2080/3080) and right now have a Quadro RTX 3000. I don’t see any significant advantage of Quadro over consumer video card.
Any advise?
On a sidenote: Anyone here has AMD Ryzen AI MAX+ 395 setup?
Short answer: No
Long answer: No, it doesn’t.
-Jakob
PS Unless you do massive renderings with a lot of hi-res textures and actually need 48GB of RAM.
Try to reduce density with cloudcompare and save as e57.
1.5GB cloud is massive for any system with Rhino.
That sounds dense for such a model, I presume you really need the accuracy.
How many points are there in that cloud?
We usually downsample to 5cm so we can work with areas like this at great speeds:
This is a 7.5M pointcloud and saves to 250MB as a Rhino file. Then we reference that file into the file we work in so we can quickly open and save the master file.
That said… I do get 120FPS on my RTX 4070 on a UWHD screen with 36 of these 7.5M point clouds in the scene (not instanced) and that is on an old intel i7-8700.
If I would need extreme accuracy then I would section the pointcloud into parts and save them separately so I could toggle on and off the visibility of the parts I didn’t need.
I don’t know if you are familiar with Unreal Engine, but they have turned 3D models into voxels for their latest nanite engine and are through that able to handle massive, massive data in realtime through clever data tree sampling. My point with that statement is just to explain that there are systems for handling massive data that Rhino doesn’t use, so throwing better hardware at Rhino won’t help much. A new graphiccard might give you 2X the fps, but I wouldn’t expect much more, and 2X isn’t enough for noticing a massive change. Far from it actually. Thus the need to be creative to figure out how you can optimize your workflow through other means than pure power.
Sorry for not giving any solutions, but if we are lucky it can help ignite a series of thoughts that might lead to something.
Just opened a 1.2GB e57 pointcloud here now and it gives me 355 fps in wireframe and 245 in rendered. (-Testmaxspeed with dynamicdraw disabled and verticalsync disabled)
This is just one cloud so not multiple, so that speeds things up.
39 million points.
Sharing this since my 4070 should be approximately the power of your 3080.
Our precision is about 1/8” or 4mm. even 40 fps is completely fine here. I was just wondering if Quadro will be needed for that one potential/future fringe case where our Pointcloud file needs to be really large. My screenshot has about 27M points and the fps is satisfactory on Quadro RTX 3000.
We also ref our PC files and layer separate setups per area and turn them on/off as we go along.
OK, Rhino defaults to the refresh rate of your monitor if it can genereate more fps than that. But if you want to check your potential fps which is quite scalable to larger pointclouds, then run -testmaxspeed and toggle the vertical sync off and see what result you get.
My experience through the last years is that the RTX gamingcards are just as fast as the equivalent gpu on a Quadro RTX card, and actually a bit faster, since Quadros are clocked a bit slower due to heat generation and stability for 24/7 loads, but with gaming being a bigger industry than cadding, this stability difference is no longer an issue. (I use the Studio drivers on all our systems though)
