ensete I am nearly 100% sure that the issue is the Windows 10 desktop rendering engine and nothing to do with hardware.
A few months ago I set up a fresh W10 install on my known good machine for testing. Same laptop/ditherig running/same driver version. W10 definitely introduces a level of strain. It didn't seem as bad as say an Apple PC, but regardless it was uncomfortable after a short period.
C_c I haven't found a computer that I can use since my old one, a laptop with 13 years, just stop working for various reasons.
Have you considered buying the same model used from eBay etc?
C_c Windows 11 is coming do you guys think it will be better or worse?
I think it will be no different to the Vista/7 update. Same OS underneath, fresh coat of paint on top (or not 😆 )
JTL We need to have real R&D here, so we can point and say to engineers at "tech companies" and such: "We've determined that the issue in question relates to LCD inversion caused by new rendering patterns as part of the display pipeline as part of OS level rendering after this build" (Okay, that was just an example I made up). My point is, if there's going to be anything to get us "out of here" it's some empirical R&D.
One approach I've thought about for some time is PCoIP. I have had strain with remote services such as Shadow PC, which turned my good machine into really bad within minutes, couldn't quite believe it tbh.
Anyways, AFAIK when you are viewing a remote machine using PCoIP, you are seeing a pixel perfect representation of that machine, warts and all. Nothing is redrawn on the client side per se, it transmits the pixel data itself to the client. This also explains why dithering fixes from Amulet exist as it saves bandwidth when dithering is off (less pixel information being sent as there are less pixel changes).
It should be possible to roll your own VMWare server and then set up various host systems (known good device, upgraded known good device, bad device etc) and compare the PCoIP information, which I believe should allow you to identify if temporal dithering is being used, or at the very least to measure if there are differences in pixel activity between OS/Drivers.
A PCoIP log viewer has been created by one of the VMWare developers, this link is a bit old but it shows some basic usage. https://www.youtube.com/watch?v=F2OAdriDCms
One of the graphs is named 'PCoIP Encoder Stats - Changed Pixels' which sounds promising.
He has a website here (not-secure) http://mindfluxinc.net/ which should provide a link to the software or to contact the author.
This doesn't fix issues with the GPU outputs or Monitor pixel inversion however IMO would provide hard data that the software itself is now making changes to the pixels that it never used to do, which is detrimental for us.