I’m preparing for the new PC build and I’m wondering what GPU type and amount of VRAM is sufficient for my workload? In most cases I use Rhino/Revit or GPU for:
basic drawing, orbiting, extruding, sectioning and creating my 3D geometry (500MB to 1GB not complex/architectural models).
99% of the time I use Wireframe or Shaded display mode (never Rendered or any other “heavy” mode) since I just care for object colors.
I don’t do any rendering or work with textures.
I plan to have 3x4K monitors (2x monitor + graphics tablet).
I know that 1), 2), 3) are mostly limited to CPU and 4) is strictly GPU. Since I don’t have a usual GPU load, I’m wondering what type of GPU and amount of VRAM will meet my needs?
My thoughts:
Since my workload is basic, I assume that GeForce will do the job. The price/performance ratio will much higher over Quadro?
I’m lost in VRAM amount. John Brock suggest 8GB/4K monitor but I assume that’s for Rendered display mode?
Hi @007
I’m running Rhino and 2x Lenovo 4K monitors on a Quadro RTX 4000, which is “just” 8GB in total - Works fine for me, even doing some light GPU-based viewport rendering (although I feel the struggle when using too many 4K and 8K textures). As always, YMMV
-Jakob
If you have a 500MB / 1GB rhino file, your GPU probably have lot of stuff to render.
CPU works more when you “manipulate” (create/edit) geometries. And here you want to have a CPU that is strong in single core tasks, you don’t need lot of weak cores.
GPU works more when you “orbit” (as you said), use clipping plane, and similar…
I’m currently working daily with 500MB files on a PC with 16GB ram and:
12 years old CPU, AMD FX 8350 … and it’s incredible how it still eats any task I throw at it with little waiting times. Wounderful CPU. Historical.
10 years old (?) GPU, a GTX 770 2GB vram. Very often the viewport freeze and lags.
If I could, I would change the GPU.
Avoid “professional” GPUs, you will not get any advantage over the commercial GPU.
Years ago I’ve seen how they flashed a different firmware on a geforce making it a quadro, and it performed better than the real thing.
Or, do whatever if money aren’t a variable in your equation.
GTX 1080 with 8 GB of VRAM served me really well from 2017 to 2021 in both Rhino 6 and Rhino 7.
The difference comes once you start rendering with your GPU. Otherwise viewport performance are more than adequate.
what, really?! i would want to try quadro drivers on my 3080. the link does not mention anything about quadro drivers on geforce, it is just about culling (omg, grasshopper influenced my use of vocabulary ) some components from the installer. i would assume that nvidia wouldn’t want to allow turning a geforce into a quadro just by installing another driver. i only believe this when i see it.
The only way to prove it is to try it yourself. Because I don’t have a GeForce card. This in case you run into unsupported hardware. Just use the NVcleaninstall mentioned above.
I have a Quadro A1000 laptop and Tesla A40 on the workstation.
Long ago in the mists of time the extra Quadro OpenGL features(that Rhino mostly doesn’t use)weren’t actually physically disabled so you could sorta turn your GTX into one with a hacked driver. That’s not the case anymore.
Also the whole “Quadro” branding is fading away, the market doesn’t care about hardware-accelerated-fake-antialiased curves anymore(THAT was the big differentiating feature!) it’s all about how many cores they have for machine learning, a Quadro driver isn’t going to magically give you more of them.
They only reason to use some other driver, is probably in case Nvidia included a new feature such as AV1 codec for OBS.
With everything mentioned, don’t expect miracles, you won’t get ECC or different GPU clock, NVlink or Stereo connector. and nothing will affect Rhino Performance.
YES, it is questionable if one would actually get any meaningful benefit from running quadro drivers on a geforce card. And YES, the quadro branding seems to be a thing of the past. There is still a distinction between gamer and pro hardware though. I think this is mainly about giving the A-model cards more vram and dropping the nv-link from the geforce cards, so you can’t just buy two cheap geforce cards to get double vram. last gen at least the 3090 had nv-link, 3080 and below didn’t. this generation, not even the 4090 has nv-link anymore.
I don’t know but I think there is still something about floating point performance that only the drivers for the quadro or A-model cards unlock on the silicon. I have no idea what benefit this would have in rhino.
No, you can’t just make stuff up and then, when someone calls you out about it, say it would be on him to disprove it. There is no trace about quadro or the newer A-model drivers running on geforce cards and the tool you linked has absolutely nothing to do with that. That tool simply cleans the driver from unwanted components.
You put out this information, so you are the one who must show some proof. Either you had done this successfully in the past or you had read an article about it somewhere. I doubt neither of the two it true.
OK, you showed you can install the geforce drivers on an A-model card. I don’t think nvidia would have any reason to prevent users from doing so. It’s a different story vice versa. Nvidia would not want users to be able to install drivers for quadro or A-model cards on geforce hardware. Though I’d love to be wrong about this.
Just stop putting out false information.
Best advice for users who use geforce cards to run cad software is to install the studio drivers.
@hitenter
No Misinformation here, Please read my above posts clearly, but I’ll summarize it for you:
1- You can install any windows Nvidia Driver (DCH one) on any GPU,
2- You may run into Driver unsupported (Likely won’t) but you can use NVcleaninstall to Get any Driver extracted and paired with your Hardware ID.
3- It won’t turn a Geforce into a Quadro, AND when It is possible doesn’t mean you should do it
I use these Techniques to install Tesla V100 on a Recycled DGX station GPU (Linux Only Drivers) and convert them into Tesla V100 and make it fully functional in Windows.
and this what I did before on my workstation running a combination of multiple GPUs (Quadro, Geforce and Tesla) … Yes I buy recycled Data Center GPUs because they work well , and I use Quadro Drivers for everything:
) some components from the installer. i would assume that nvidia wouldn’t want to allow turning a geforce into a quadro just by installing another driver. i only believe this when i see it.
