cancel
Showing results for 
Search instead for 
Did you mean: 

Drivers & Software

Anonymous
Not applicable

Custom Resolution not accepted despite copying from CRU

Hello,

On my new laptop (E495 with Ryzen 5 / Vega 8) Windows 10 won't let me select the correct resolution for my cheap 2k LCD panel. The panel is in portrait mode 1440x2560 (yes, vertical) with a refresh rate of 49.999Hz and should thus be theoretically supported by the laptop (at the very least HDMI 1.4b, if not 2.0).

It (or rather the LCD driver board behind it) does not support downscaling or anything fancy, only it's native resolution.

It has worked on my old PC /w Nvidia GPU before on Windows 10 and still works on that PC with Linux (don't have Windows 10 on it anymore).

Unfortunately, windows only detects the LCD as 1280x1024 with no other option, resulting in the LCD board rejecting the signal (blue blinking instead of green light, as I said it needs exact parameters or it does not work at all).

I've tried reinstalling the GPU drivers and display drivers to no avail. 
I tried the CustomResolutionUtility and it automatically detects the proper resolution just fine (as the only supported resolution as expected, in the detailed resolution area):

LCD_CRU.PNG

Unfortunately messing around with that tool didn't actually fix anything, windows still only let me select 1280x1024.

So I tried setting that as a custom resolution in the radeon software settings by copying the parameters:
LCD_AMD.PNG

leading into an error "Custom resolutions settings are not compatible with this display" (and yes I picked the correct display).

(Note: Presentation let's me chose between Progressive and Interlaced so it corresponds to the interlaced toggle in CRU. Also tried Refresh rate of 49.999 with no difference)

This is what Radeon software thinks the display supports:

LCD_AMD_Specs.PNG

Some observations that might be part of the problem:

The refresh rate of the display is apparently 49.999. Don't know how sensitive it is here, but when I select 49.999Hz with Timing standard any of (CVT, CVT Reduced Blankind, GTF) it does accept but cuts the refresh rate to 49Hz. But then (with 49Hz/50Hz and GTF) I do observe the LED flashing green for some 2 5-6s intervals and then off for a while. Not done the math but maybe that indicates that the refresh rate is just slightly off and at some point it manages to read the signal correctly but then gets out of sync.

What's up with this disparity in specs and how can I fix it? Did I enter the custom settings wrong?

Again, the LCD is fine (albeit probably not properly supported). It works in Linux out of the box, without even a hitch.

Thanks for any insights into this topic!

0 Likes
3 Replies
Anonymous
Not applicable

Here's the output from xrandr --verbose on the linux rig that DOES correctly recognize the LCD out of the box:

LCDPanel_Manajro_xrandr_verbose.png

Which also shows that the timings CRU detects are indeed correct (afaict).

0 Likes
Anonymous
Not applicable

Ok this has never happened in the past few days I've been trying to get it to work but now for the first time ever, after plugging it in again after testing it on linux, windows did indeed pixk up the correct resolution, as did AMD Radeon. Here are the default values as detected by the drivers:

LCD_AMD_new.PNG

However, the timings were wrong, and thus, despite the LCD showing green light and attempting to display the signal, it's mostly black and some occasional flashes of distorted image. However, I still could not get AMD Radeon software to accept the correct timings - what did I do wrong?

0 Likes
Anonymous
Not applicable

And now, without any change to the settings, it does show a full image, but it has severe vertical artifacts that flicker (each line seems to jump back and forth between two pixel columns),

LCD_Glitches.PNG

I have no clue what these glitches mean (as in, what setting is wrong that could cause that). Also sorry for the image quality, is shot through lenses.

Polarity is weird, can't set it to negative negative as CRU implies, but either +/- and -/+ produce the same result.

Timing is still completely wrong but it seems to work for some reason... no idea.

Also sorry for mass posting, was not my intention to suddenly get new results right after posting...

0 Likes