Tegra Orin NX add new custom display resolution

Hi All,
We have a splicing screen with various combinations of special resolutions.(25841462p60, 25761462p60) Currently, with our Orin NX module (R35.4.1)in DP mode outputting the screen, we find that through xrandr and NVIDIA settings, only some standard resolutions are supported. When we plug in the splicing screen, the ORIN NX GPU only provides us with the option of 4KP60, which causes occasional flickering on the display of the splicing screen.
How can we add support for some non-standard resolutions?
we have tried these suggestions in Google but it seems reject via NV driver.
https://wiki.archlinux.org/title/Xrandr#Troubleshooting

nvidia@yuan-6n0snx:~$ cvt 2584 1462
# 2584x1462 59.95 Hz (CVT) hsync: 90.83 kHz; pclk: 319.00 MHz
Modeline "2584x1462_60.00"  319.00  2584 2776 3048 3512  1462 1465 1475 1515 -hsync +vsync
nvidia@yuan-6n0snx:~$ cvt -r 2584 1462
# 2584x1462 59.97 Hz (CVT) hsync: 90.20 kHz; pclk: 247.50 MHz
Modeline "2584x1462R"  247.50  2584 2632 2664 2744  1462 1465 1475 1504 +hsync -vsync
nvidia@yuan-6n0snx:~$ sudo xrandr --newmode "2584x1462R"  247.50  2584 2632 2664 2744  1462 1465 1475 1504 +hsync -vsync
[sudo] password for nvidia: 
Sorry, try again.
[sudo] password for nvidia: 
nvidia@yuan-6n0snx:~$ 
nvidia@yuan-6n0snx:~$ xrandr --addmode DP-0.2 2584x1462R
X Error of failed request:  BadMatch (invalid parameter attributes)
  Major opcode of failed request:  140 (RANDR)
  Minor opcode of failed request:  18 (RRAddOutputMode)
  Serial number of failed request:  27
  Current serial number in output stream:  28

and we have tried modify /etc/X11/xrog.conf

add an option in Section “Device”
Option "ModeValidation" "AllowNonEdidModes"

Hello,

Your topic will be best served in the Jetson category.

I will move this post over for visibility.

Cheers,
Tom

Hi,

We only support to read customized EDID from xorg.

See - Appendix B. X Config Options :

Option "CustomEDID" "string"

Thanks for the reply,
Can this setting handle output resolutions that are not multiples of 16? Based on previous experience, NVIDIA’s buffer management requires multiples of 16.

More precisely… what NVIDIA driver supports are mostly the CEA modes.