AMD RV770 refresh -> RV790

[Blazkowicz]

For the approximatively 5 people that use Radeons in Linux (out of the 200 people using Linux in a way that warrants a high-performance GPU driver)? I sincerely doubt it since FGLRX has been hardly loved in those spaces, and the true path is FOSS drivers 111oneoneone, which will keep on being developed. This is just formalising something that has been going on inside for a while, IMHO. The alternative would've been to keep quiet and just keep the DeviceIDs in the inf file - would that have been better?

you're mostly right
but if I were to buy an AMD netbook or laptop, with only 1 GHz bobcat cores, I would be pissed to run a slow driver. bad for casual lan gaming and wastes CPU cycles, that the browser already eats. no video decoders either to save on battery life. even some heavy h264 video playback may be compromised.

 

[AlexV]

you're mostly right
but if I were to buy an AMD netbook or laptop, with only 1 GHz bobcat cores, I would be pissed to run a slow driver. bad for casual lan gaming and wastes CPU cycles, that the browser already eats. no video decoders either to save on battery life. even some heavy h264 video playback may be compromised.

But Bobcat is unaffected by this, and by the time it ends up in the same bin it will probably FGLRX would've gotten as good for it as possible. I also don't think that ATI manages to do accelerated video decode in any worthwhile capacity under Linux (yeah there's XVBA, but that seems hardly great and not quite supported). Prior cards also didn't support CL (bar the RV7xx ones, but I think that's capped at 1.0), which could've been argued as another reason to have FGLRX.

 

[CarstenS]

Agreed, but it's IMHO too late for that now sadly, they've painted themselves into a corner with the monthly thing. I think that doing away with some of the validation and overhead involved with maintaining a monthly release cadence would help with the overall quality of things.

To this, I have to agree in turn.