[d3d8/9] Backport several d3d8 changes and SM0 support

[WinterSnowfall]

Now that you've gotten a new stable release out and there's less risk for things to blow up 😉, I've taken the liberty to backport several other retro-friendly changes and added features from dxvk, namely:

• d3d8 config options are now documented part of dxvk.conf
• several small d3d8 fixes and improvements
• proper support for a "Shader Model 0" mode, which essentially force exposes no programmable shader support. Note that some games will NOT like this one bit, however most games from the d3d8 era and some from the early d3d9 era will in fact provide a fallback path in which they will only use fixed function shaders (aka T&L, which is all that d3d7-capable cards supported at the time). This can be used for nostalgia reasons, to force games on simpler graphical paths that they would have used on limited capability hardware.

I've only tested the above SM0 mode briefly, but it seems to work just fine:

Don't worry, BloodRayne is one of those games that doesn't react well to a lack of PS 1.x support, so it looking like that only confirms the logic is working properly, and this is how it would have looked even back in the day on, say, a GeForce 4 MX card, if anyone had tried to play it. I've also confirmed it looks perfectly fine without the option.

[pythonlover02]

Thanks for the help, mate. i will test it locally and probably merge it tomorrow. As compiling only one branch of DXVK-Sarek takes half an hour on my system, so an hour total, plus testing the changes on both branches.

[WinterSnowfall]

No worries, take your time. That being said there isn't anything in there I'm particularly worried about, so it should be good to go.

As compiling only one branch of DXVK-Sarek takes half an hour on my system, so an hour total, plus testing the changes on both branches.

Ouch, that sounds painful. You might want to consider setting up ccache on that system, so that only actual changes get recomplied - it should help speed up things quite a lot once the cache is populated on the first run.

[pythonlover02]

everything work great on my end :)

[pythonlover02]

Ouch, that sounds painful. You might want to consider setting up ccache on that system, so that only actual changes get recomplied - it should help speed up things quite a lot once the cache is populated on the first run.

Yeah, it is. I should probably do that. About two months ago, I had my main PC working, which had a Ryzen 7. It compiled the project without any issues, . But, for whatever reason, the BIOS killed itself and damaged some components, so I have to use my backup/testing PC, witch its an okey PC, but not great by any means.