Rotating/panning the viewport on my laptop has been quick since I had encountered slow display a few months ago.
Today, I hooked up to a large 2nd display via the display port and experienced lag mostly in when the perspective view is maximised, or quite large. When in multiple viewport mode, seems ok.
Vendor Name: NVIDIA Corporation
Render version: 4.6
Shading Language: 4.60 NVIDIA
Driver Date: NA
Driver Version: NA
Maximum Texture size: 16384 x 16384
Z-Buffer depth: 24 bits
Maximum Viewport size: 16384 x 16384
Total Video Memory: 2 GB
I’ve updated the Quadro drivers, and Rhino to v6.2.
I’m still experiencing lag when connected to a second display in the ‘Extend’ display mode. From a Dell Precision M3800 laptop with a 4K display and Quadro K1100M to a Dell monitor via the HMDI port.
The models are instant on the laptop, but when moving the Rhino working window upto the 2nd display it significantly slows…
I’ve just tested exactly the same thing in v5, and the issue is not there.
It’s a bit frustrating that this hasn’t been resolved in the full release of v6.
Verify that Rhino is seeing your Quadro GPU in Help -> System Information...
Click and drag the Rhino window over to the external display.
Close Rhino by going to File -> Exit.
Now start Rhino again. Does it open on the external display? If so, check Help -> System Information... again. Does Rhino still use the Quadro or some other GPU?
To note that the main laptop display is 3840x2160 and the 2nd is 1920x1200.
The problem doesn’t exist when using the ‘Duplicate’ mode where both
displays are showing the same. Although the 2nd display in this case seems
to reconfigure it’s resolution. Things get a bit bigger and seem to be
lower DPI.
Also, after following David’s steps where Rhino starts up on the external monitor…were things slow then? Or do you have to move Rhino over to the laptop before things get slow again?
This all sounds like configuration issues, none of which sounds like Rhino can know or do anything about…but there may be something you can do on your end with the configuration settings and the NV Control Panel, we’ll have to see if we can reproduce this situation in the lab and then see if anything can be done about it from a configuration point of view…but I’m doubtful that there’s anything Rhino will be able to do.
I think that’s indicative of a resource issue…V5 uses probably around 1% of your GPU’s resources… V6 will use 100% of your GPU’s resources if it needs to. Just to give an example…
Your laptop is running at 3840 x 2160… which means just the screen alone requires a minimum of 32MB of video memory just to store what you’re seeing on your desktop. Each viewport in Rhino V6 is managed by a Multi-sampled frame buffer and a single frame buffer… A frame buffer consists of a 32bit RGBA buffer and a 32bit floating point Depth and Stencil buffer…and take note what I said above…there are two frame buffers per viewport (a sampled one (AA) and a non-sampled one).
Given that, let’s make things simple to work with… Suppose when Rhino starts up it uses the screen real estate uniformly … So a 4 view layout means that you would have 4 views each running at 1920x1080… So lets break down the memory usage…
1 Viewport would be:
1 Sampled Color Buffer at 1980 x 1080 x 4 x AA … so for 8x AA that’s 1980x1080x32 = ~65MB
1 Sampled floating point Depth/Stencil buffer at 1980x1080 x 4 x AA … again, at 8x that’s ~65MB
1 Non-sampled Color buffer at 1980 x 1080 x 4 = ~8MB
1 Non-sampled floating point Depth/Stencil buffer at 1980x1080x4 = ~8MB
…for a grand total of 65 + 65 + 8 + 8 = ~145MB …just in buffers alone
Since there are 4 viewports, that’s 4 x 145, which is roughly 580MB of video memory that must be managed contiguous memory by the drivers… That’s over 25% of your current VRam that you currently have…and that’s just talking about the frame buffers… There’s also the screen buffer I mentioned, which is another 32MB, and now thrown in your second monitor, which now also requires a block of contiguous memory just to manage what you see on it …
Rhino V6 then also starts allocating GPU memory for almost every thing you see in your model… So things can start adding up quickly… Now, I realize that so far I’ve only accounted for about 35% of your VRam…but you asked why V5 doesn’t show this problem…and I’m guessing because V5 doesn’t instantly use up 25% of your video memory just to start up and create a 4View layout.
Whether this is actually the reason is purely speculation on my part…
-Jeff
P.S. How the drivers choose to manage the memory once it becomes largely blocked like this is also a factor, and an unknown on my part.
I doubt there’s much they can do, as Jeff says the drivers have a big say on how the memory gets managed, and 2 GB just isn’t a lot of video memory for 3D these days–I used to have 2 GTX 970s and the limiting factor on their image quality in SLI games was that they only had 4 GB of memory, and I’m not at 4K resolution.
People have been asking since the days when Matrox was a name in 3D video cards for Rhino to make more extensive use of the hardware, now that it does…well it’s going to make more demands on your hardware.