this is a known issue in jetpack 5.X as well. confirmed on an xavier nx devkit running 5.0.1.
this is not jetson specific either… it is an issue with all nvidia gpus running xorg, there have been many reports in the nvidia forums over the years. it is just less and less noticeable the faster the cpu is in the running system.
How this isn’t top priority above all other issues is beyond me. especially since nvidia keeps pushing to keep using xorg even on modern linux distros
I’d also like to point out that the original issue was opened over a year ago… And the public was told a fix was released.
This seems ridiculous–promising a fix, pretending a fix is released, not actually fixing the issue, and then saying you have to buy one of our newer (all sold out or upsold at ungodly prices I might add) boards and trust us that we’re actually going to fix it this time in order for basic functionality to work.
For using Jetpack 4+ Jetson Nano, please not to use this mode.
Seriously? X11 is a fundamental part of using a graphical environment on these devices, especially as Wayland has its own issues on them. “Just don’t rotate the screen with X11” seems like a tall request given prod modules are marketed as having a working Linux OS and clients of yours are actively using these modules in a capacity which requires X11 rotation for the device to be viable for usage.
Either add t210 support to JetPack 5.x or give us a real fix for this issue that was supposed to be fixed ages ago.
also just incase anyone forgot how terrible it is… here is a screenshot comparison
What is the recommended course of action for Jetson Nano users to avoid this critical bug that we were told was already fixed?
On Jetson Nano, please run in default landscape mode and not enable ForceFullCompositionPipeline.
Just want confirm this. Is it possible to use EGL app instead of GLUT or SDL?
all applications, the desktop, and everything rendered to the display is affected. It does not matter if it is GLES, EGL, opengl, vulkan, SDL2 (using opengl), etc. everything passes through your broken composition pipeline and is affected. the developers know this.
the application choice is purely for reference so that you can easily observe the issue. the framepacing issue happens on everything.