Hi there, I would like to do projection warping from a laptop. I have a laptop that has a 3070RTX mobile GPU.
My goal is to fix the warp on a curved monitor. Running the WarpBlendSample ends with each demo saying that the scanout functions are not supported.
Is there a laptop that runs a version of RTX that does support GPU scanout calls? I would prefer to stay in the laptop space and not go desktop. However, if I do need to go to desktop, what is the minimum RTX GPU card that I would need?
Also, it would be nice to know exactly what this functionality does. Is this something that I can do outside of my applications? For example, my application that I want to fix is a Unity app, so can I make these calls outside of Unity in another app to warp the screen? Even better, can I make these calls within a Unity C++ dll so I can control the warping in real-time?
Thank you,
Peter
I am in the same position. I have a new Dell Precision 7780 with an A4000 Ada Gen GPU. Every time I try to run the warping sample, I get the an “NVAPI_NOT_SUPPORTED” error. This worked just fine on my older 7550 laptop. The A4000 is truly a “workstation” class GPU. Is there a reason NVAPI is showing the feature as unsupported on the newer card?
I believe I have found the culprit (at least for my case). I enabled “Switchable Graphics” in my BIOS which is supposed to allow Windows to select between the on-chip graphics and the RTX. My logic was that I wanted the laptop display to be handled by onboard graphics while freeing the RTX to only handle the external projectors. It turns out that this setting in the BIOS also seems to affect NVAPI. I’ve tested this on two computers, one with an RTX4000 and the other an RTX4000 ada. Both began working as soon as I disabled the switchable graphics. I fought this for 3 days before I figured it out. Hope this saves someone else the time.
Wharp’n’blend is a scanout feature of the professional Nvidia GPUs like your 4000ada.
We can only apply it, when we drive a screen directly! Normal operation for the mix of CPU-integrated GPU and discrete NV GPU is for demanding workloads to be rendered by NV, framebuffer content to be copied into system memory, for the integrated GPU to scan out to screen.
Some notebooks can’t have the discrete NV GPU ever scanout directly, some have the NV GPU wired to only a specific output connector of the device, some/few have “switchable gfx”, where an additional HW MUX’er is required to switch the signals between either GPU and screen(s).
If no screen is directly connected to the NV GPU, it correctly reports it cannot support the scanout/warping feature, since it is NOT scanning out any pixel…
A more descriptive/helpful error message would be vendor specific, and several levels higher to implement than just the NVAPI level - apologies for not being able to give better hints with the NVAPI error message…
-Frank