
Originally Posted by
bridgman
re: "without HyperZ" -- yes, that is just one more step in ongoing performance optimization, not a fundamental change in what you can do with the driver. It is important, but IMO it doesn't define the line between "support" and "no support".
re: Steam, I think the initial focus for Steam is via the Catalyst driver, not sure what the plans are re: getting games running on the open source driver. That was probably the right decision at least until we have consistent launch-time support for new GPU generations. We're still aiming for that to start with the Sea Islands generation.
Re: transitioning support, the decision to move a GPU generation to legacy is made based on market-weighted requirements across all the OSes supported by the Catalyst code -- the Linux Catalyst driver basically "looses the ability to leech off the work we do for other OSes" if you want to use forum terminology.
When we hear about a transition coming we do look for specific things we can do on the open source side to minimize the impact (eg the initial power management work Alex pushed out when we heard about the r3xx-r5xx transition), but since Catalyst Linux exists primarily to take advantage of code sharing across multiple OSes the decisions are made at an "all OS" level not at a Linux-only level.
If Catalyst Linux were a Linux-only code base then the answer would obviously be "yes we would coordinate", but it would also be open source and have pretty much the functionality of the current open source stack, except for the fact that work would have started a few years earlier (which would help). The problem with that approach is that it would not have given us a suffficiently capable 3D workstation driver, which in turn would have meant that the funding for Linux graphics driver work probably would never have materialized in the first place.