RTX 3060 and monitor pixel count

I think this may be a question for @John_Brock , as I’ve seen several comments made by John about the relationship between video card memory and the number of monitors and their pixel count.

I’ll be adding a fourth monitor to my setup shortly that I can use, amongst other things, for Grasshopper without blocking my view of my standard Rhino viewport layout. The RTX 3060 seems peculiar in having a large (12GB) onboard memory. Not sure why? I’ve not seen a 3070, 3080 or even a 3060Ti with as large a memory. Does this make it well suited to supporting a high pixel count? The monitors are one 2560 x 1600, two 1600 x 1200 in portrait format and a 1920 x 1200 touchscreen.

I don’t do much in the way of rendering so I don’t think I need a bleeding edge GPU. Of more concern is cool/quiet running, energy efficiency and of course the ability to manipulate models smoothly while modelling. I’ll also be making the jump to Windows 11 when I add the fourth monitor. From one or two comments made by others here, that in itself may give me a few headaches. My current GPU is a RTX 2070.

Possibly - I have two 4K monitors here and in my experience multi monitor support is abysmal on Win 11 - mainly concerning what happens when the computer sleeps and then wakes up, the window sizes and monitor placements are not remembered. I also have a 3060 with 12Gb. I’ll be curious how it works for you with four monitors.

That 4-GB-per-screen recommendation is pretty conservative, I suspect an awful lot of people aren’t meeting that, and all together you don’t have as many pixels as 2 4K screens.

The minimum recommendations for GPU VRAM from the developers are 4GB VRAM for each standard resolution monitor, and 8GB VRAM for each high resolution monitor.

12GB should be adequate for three standard resolution screens.
32GB should be adequate for four high resolution screens.

John - What would count as a “standard resolution monitor” in this context? 1920 x 1200? A VRAM Gb / monitor megapixel ratio would be a constructive yardstick.

Yes, but nobody has ever told us what type of usage these ‘recommendations’ are based on or what math was used to calculate these values. Already the ‘math’ stated above is inconsistent. A 4K “high” resolution monitor has 4x as many pixels as does a “standard” 2K or HD resolution, but the VRAM recommendation only doubles…

And it’s obvious that these are not minimum requirements, many users have reported that Rhino ‘runs fine’ with far less than the recommended minimums. For example Rhino 7 still runs adequately (no crashes, no major display glitches) on a couple of my old ‘emergency’ laptops that I occasionally have to lend out to students if their computer goes haywire. And these 10 year old devices have 2Gb video cards driving HD screens (i.e. half the recommended VRAM and very old cards).

And I should need 16 Gb for my two 4K monitors, but I only have 12. Rhino runs fine for “normal” use. Of course if I put one Rhino instance full screen on each monitor and run Turntable on both files with 1000 spheres in shaded mode, everything does slow to a crawl. But that is pretty extreme.

1 Like

I recall @jeff explaining the calculations on at least one occasion in a Forum thread. I do not have it bookmarked.
Maybe he has it handy or can reprise it when he comes up for air.

I run two 27". 2550 x 1440 “standard resolution” monitors with 6GB of VRAM, with the same recommended Windows scaling on both. On occasion, I can see a pause as the GPU juggles memory to paint both screens.
For me, it’s fine.

I have 4GB of VRAM on my Win 11 Alienware x14 running a GF 3050Ti. I only run the built in laptop screen so it runs very quickly and well.