I'm not really worried about Windows OpenGL performance, even the new Minecraft edition (which performs enormously better all around) uses DirectX. Browsers use ANGLE to convert GL to DirectX for a reason, it's a crap API all around on Windows (not that it had to be but that's how it is).
The driver bugs are what I'm most worried about though. I remember when I got a 4670 in 2008 there was a cursor corruption bug (google it, it's common) and if you didn't shake the mouse enough for it to fix itself it'd crash. Then I built a new PC with a 5770 and I still had it. Later a new PC with a 6950 and still had it. Then 2x280X and still had it. It even occurred on Linux once. Then I went Nvidia for a few generations and it seems AMD finally fixed it. It seems fixed now (maybe because of new hardware) based on internet searches but from 2008-2015 the bug was there. And this is just one example of things.
Not that Nvidia drivers have ever been perfect but I never ran into this kind of shit so often for so long on their cards. Hopefully I don't end up regretting the 6900 XT because of it.
This cursor corruption bug was ridiculous. I ended up not playing some games (I think Dota) because I would get this bug. I went with an Nvidia card as my next one because of that as well.
It's interesting how something that seems relatively minor could have such an impact on purchasing decisions.
This used to be a pretty common bug, I think. The cursor is often drawn using a special "hardware mouse cursor" feature specifically designed to make a tiny bitmap move around the screen as efficiently as possible. It'd be easy for a driver to mess that up and cause the cursor -- and ONLY the cursor -- to look garbled.
You can see problems today like the mouse cursor not respecting Windows 10's night-shift color filter for the same reason.
On a thread yesterday I said that Windows' settings and lack of native keyboard remapping (without registry editing) were the most ridiculous things about it.
This pointer not respecting night-shift hits the trifecta hat trick. There's a registry fix, of course, but the first time I saw it, I thought "huh, that's weird". How did this get out without somebody on the release train thinking to themselves the exact same thing?
Cool, I never heard of that hardware mouse cursor feature!
Makes sense to optimize that though as people will usually be updating the mouse position frequently
The driver bugs are what I'm most worried about though. I remember when I got a 4670 in 2008 there was a cursor corruption bug (google it, it's common) and if you didn't shake the mouse enough for it to fix itself it'd crash. Then I built a new PC with a 5770 and I still had it. Later a new PC with a 6950 and still had it. Then 2x280X and still had it. It even occurred on Linux once. Then I went Nvidia for a few generations and it seems AMD finally fixed it. It seems fixed now (maybe because of new hardware) based on internet searches but from 2008-2015 the bug was there. And this is just one example of things.
Not that Nvidia drivers have ever been perfect but I never ran into this kind of shit so often for so long on their cards. Hopefully I don't end up regretting the 6900 XT because of it.