02:51 airlied: imirkin: what codecs does the nouveau decoder in mesa support?
02:51 airlied: and it is just decode?
02:51 imirkin: just decode
02:51 imirkin: depends on the gen
02:51 airlied: any h265?
02:51 airlied: or just h264?
02:51 imirkin: no h265
02:51 imirkin: i think that's maxwell+ -- i tapped out by then
02:52 imirkin: airlied: https://nouveau.freedesktop.org/VideoAcceleration.html
02:52 imirkin: vp5 = kepler
02:53 airlied: imirkin: okay, maybe you can glance at 15258
02:53 airlied:has checked it builds yet though
02:54 imirkin: sec...
02:54 imirkin: airlied: to be clear, you're singling out H.264 as opposed to other MPEG-4 videos?
02:55 airlied: yes h264 and h265, whatever mpeg-la has current patents on
02:55 imirkin: airlied: also, are you _sure_ you compiled?
02:55 airlied: nope I didn't compile :-)
02:55 imirkin: oh
02:55 airlied: just going to let CI do it now
02:55 imirkin: well, there's a typo in nv84_video
02:55 imirkin: presnet -> present
02:56 airlied: that's more embarrassing that I'd expected
02:57 airlied: I suppose I could add mpeg4 to it, I just don't know if there are any real patents left there
02:57 imirkin: not for me to say
02:57 airlied: yeah nor me :)
02:57 imirkin: there's also VC-1 - no clue about the patent situation there
02:57 imirkin: but it's also not exactly the most popular
02:58 airlied: oh I should dig into vc1 a little
02:59 imirkin: mpeg1/2 is hopefully all expired by now
03:00 imirkin: (but again, don't get your legal advice from me...)
03:00 airlied:definitely is not looking for legal advice :-P
03:00 airlied:has had quite enough fun learning about the crap show that is video codec licensing
03:00 imirkin: alllso - maybe more prudent to do it in st rather than individual drivers? but i guess there could be differences there...
03:01 airlied: imirkin: I'd rather do it in the drivers
03:01 imirkin: ok. we controlled s3tc in mesa directly, not the backends
03:01 imirkin: which was a comparable situation
03:01 airlied: that way some vendor could actually say they covered all impls of their hw and we could handle it
03:01 airlied: well there are 3-4 st
03:01 imirkin: i guess yea
03:01 airlied: so it's nearly as many as the backends :P
03:01 imirkin: crocus soon too?
03:02 airlied: crocus video might be able to land once I get this in :-)
03:04 airlied: for vulkan it would have to be in the drivers anyways
03:05 imirkin: i guess yes
03:05 imirkin: yeah, it was just a suggestion. i'm not in any way set on it
03:11 airlied: okay I added vc1, seems like a good plan
03:13 airlied: next step is to convince people who work for hw manufacturers that their company doesn't pay the patent license :-P
03:15 imirkin: airlied: well, the nouveau bits seem right now, assuming those defines get set as expected.
03:16 imirkin: whether the whole thing is a reasonable idea, i have no comment
03:18 airlied: but armchair lawyering is such fun :-P
03:18 imirkin: heh
03:18 imirkin: happy to armchair lawyer based on how the laws SHOULD work
03:32 airlied: okay tested the defines work on radeonsi at least :-P
18:11 mynacol: 6piz7wk: I bet you didn't register with OFTC, so they don't see your messages
18:11 mynacol: And AFAIK nouveau has no support for vulkan. Period.
18:12 Lyude: it's a WIP, there's no work on it in the kernel tree yet though
18:12 Lyude: there's some other stuff that needs to be done on nouveau first before we'll be able to support vulkan that's in progress as well