Originally posted by JoshuaAshton
View Post
I didn't mean to say that the code itself gets contributed upstream. but that there are a small number upstream commits are based on code from vkd3d-proton and that some devs who work on vkd3d-proton also occasionally contribute upstream. I also recall seeing other commits similar to what get merged in vkd3d-proton, but I supposed I was mistaken and was correlating two separate things. If I am wrong about the scale of cooperation I will gladly correct what I said. the combination the three made the impression that there was closer workings then there actually is and is what I meant by "work gets up streamed".
I didn't mean to imply that game specific fixes are a bad thing. Mesa does game specific fixes, and looking at change logs it's evident that AMD and Nvidia do as well for their windows counterparts. working around game bugs is a critical part of the job, If hacks aren't the right term for it I am not sure what the correct term would be. for me a hack has always been a specific change in the general program to do something else. I don't think that making a game work should be on the driver devs. to my group of peers, game specific changes, Kernel quirks, etc. would all be "hacks" designed to make up for the shortcomings of something else.
I also meant game related... changes that for instance, may reduce performance in more specialized applications. or apps that are broken that they don't benefit from fixing but upstream does. I don't necessarily mean d3d12 "compute" would be off the table. (DirectML for instance is something that could vkd3d-proton could supposedly benefit from though i don't know of any games that uses it but MS does advertise it). things like VEAI, I wouldn't try to use proton for, but rather upstream is what I meant.
again, you have my apologies, if you can elaborate on the corrections to be made I will gladly edit my post to reflect them, if not ill just point towards your comment.
Leave a comment: