03:20 Ein: we may be retiring an old cuda cluster, has GTX 470s in it.
03:21 Ein: if anyone wants one, ping me and i'll try and get it to you
06:26 AndrewR: I think https://gitlab.freedesktop.org/xorg/driver/xf86-video-nouveau/-/issues/385 can be solved after some variant of https://lists.freedesktop.org/archives/nouveau/2020-August/036472.html lands ?
20:24 AndrewR: imirkin, hello and welcome back
20:24 imirkin: hi
20:29 AndrewR: imirkin, looking at not updated xf86-video-nouveau you still had no time to test your patch ?
20:29 imirkin: i'm running it now
20:29 imirkin: waiting to see if issues arise.
20:29 AndrewR: imirkin, ah, thanks!
20:32 AndrewR: imirkin, any idea how to debug further my "vp4-sometimes-hang-with-prime-on-mplayer-quit, but not when I enable NOUVEAU_VP3_DEBUG_FENCE 1" ?
20:32 imirkin: not really
20:33 AndrewR: imirkin, I tried to disable bsp/vp fences but my lack of C understanding prevented my attempts from working. I think this debug option also introduces assertion if I try to decode mpeg2 via vaapi via gstreamer .....
20:34 AndrewR: imirkin, also, I think I saw some 'motion artefacts' in h264 trailer (downloaded from youtube), want to look at screen capture?
20:34 imirkin: nope
20:34 imirkin: i know what they look like :)
20:34 imirkin: it happens with h264 across all gens
20:34 imirkin: don't precisely know what we're doing wrong
20:35 AndrewR: imirkin, but there is no way to intercept bsp/vp/ppp communication while they run on binary nvidia driver?
20:36 imirkin: you can intercept what commands / data are sent to individual engines
20:36 imirkin: can't really look at the inter-engine data
20:36 imirkin: (but it doesn't really matter anyways
20:36 imirkin: if you mean "why haven't you figured out why the h264 corruption appears"
20:36 imirkin: then the answer is that i just suck at computers, coz i tried and failed.
20:37 AndrewR: imirkin, you at least fixed a lot of issues and develop a lot of useful paches ....