I checked already if it depends on fractional scaling, but that´s not the case. Will double check though. 
My CPU is an AMD 9700X and i´m using its iGPU. I´m also running a very recent MESA version (25.x, I have to check the exact version)
and I also just found this thread: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12809 where also Emacs is mentioned. 
Maybe it´s related. 

Am Mo., 25. Aug. 2025 um 09:28 Uhr schrieb Po Lu <luangruo@yahoo.com>:
Eli Zaretskii <eliz@gnu.org> writes:

> [Please use Reply All to reply, to keep the bug tracker on the CC list.]
>
>> Date: Sun, 24 Aug 2025 21:07:05 +0200
>> From: Marco <marconeumaier@gmail.com>
>>
>> Hi,
>>
>> thanks for looking into it. I made some more tests today.
>>
>> Findings:
>>
>> The issue occurs only in a Wayland session.
>>
>> When running Emacs in a true X11 session, the problem does not appear at
>> all. If I'm wrong I'll correct myself.
>>
>> Setting GDK_BACKEND=x11 while under Wayland does not eliminate the
>> problem, which suggests it still goes through XWayland and the
>> compositor path.
>>
>> The artifacts are visible in screenshots and occur sometimes across the
>> whole display width, though their colors sometimes differ slightly,
>> which suggests this is a rendering bug.
>>
>> Most of the time they look similar to what you see here:
>> https://lists.gnu.org/archive/html/bug-gnu-emacs/2012-05/msg00023.html
>> or
>> https://lists.gnu.org/archive/html/bug-gnu-emacs/2018-01/png1wmnm51YOW.png
>>
>> It only seem to occur when the emacs window is maximized!
>>
>> With absolute line numbers, I cannot reproduce the problem — it only
>> appears with relative line numbers.
>>
>> Conclusion:
>> This looks like a rendering bug in the GTK/Wayland (or XWayland) stack,
>> but Emacs may be triggering it in a specific way when drawing relative
>> line numbers.
>>
>> This is maybe related: When opening emacs in nw mode and using relative
>> line number I also see artifacts, i.e.  lines across the whole window
>> width in a color slightly darker than the background. The height of the
>> line is the same like the block cursor height.
>>
>> Again: This does not happen in a real x11 session.
>
> Thanks for the footwork.
>
> Po Lu, any ideas?  Are problems like this with Wayland known to exist?

I'm not aware of any such problems, but I would still ask the OP whether
he has enabled fractional scaling on his Wayland desktop, which is a
recent innovation that has the potential to produce artifacts in
rendered text (particularly if the compositor believes a frame to be
completely opaque).