00:15 borrougagnou: oops bug
05:30 mupuf: Lyude: shit, I forgot about this! I think I broke my instance of gitolite
05:30 mupuf: I'll try to fix it up during the weekend
05:31 mupuf: and I need to fix up wtrpm alos
05:32 mupuf: actually, how about using gitlab now?
05:32 mupuf: this way we can stop relying on my server for this
05:57 OverSampled: Hello
05:58 OverSampled: I am trying to get my Quadro K1000M to run UHD@60hz over a DisplayPort to Hdmi 2.0 adapter
05:58 OverSampled: I was successfully able to do so in windows with up-to-date drivers there
06:00 OverSampled: I found references to this fix on the proprietary drivers, which seems to work for some people (at least on other cards): https://www.nvidia.com/Download/driverResults.aspx/126577/en-us
06:01 OverSampled: In it, they fixed the max pixel clock for Hdmi 2.0 adapters
08:26 sigod: does vdpau make a big difference in performance for watching videos and ensure no audio sync problems?
17:48 AndrewR: hello, all! Today Cinelerra-GG (video editor) gained ability to use vaapi/vdpau for decoding some videos. I tried to test it with mpeg2 video, but run into issue of too low video decoder clocks on nv92. Video decoding with mplayer was fater with NOUVEAU_PMPEG=1 (fixed-function mpeg2 decoder block, IIRC), but guess this env. var not wired correctly for VA-API mesa state tracker + nouveau ? At least if I run vainfo with NOUVEAU_PMPEG=1 any decoder entries
17:48 AndrewR: disappear ....
19:04 jcdutton: Hi. Has anyone managed to get nouveau working with kvm gpu passthru
20:09 karolherbst: jcdutton: huh? I don't see why it shouldn't just work
20:09 karolherbst: or are there issues you run into?
21:04 jcdutton: karolherbst, The old nvidia graphics card works fine with nouveau on the host. If I do passthru on a kvm guest, the nouveau driver fails to at the probe stage.
21:04 karolherbst: hum
21:04 karolherbst: weird
21:04 karolherbst: but the device is detected?
21:05 karolherbst: what is it that nouveau is complaining about?
21:19 jcdutton: karolherbst, any ideas ?
21:20 karolherbst: not without a dmesg at least
21:20 karolherbst: what is nouveau complaining about?
21:20 karolherbst: maybe the device wasn't correctly passed in?
21:21 jcdutton: http://paste.debian.net/1079639/
21:23 karolherbst: jcdutton: huh? this is a weird kernel
21:23 karolherbst: ohh, no
21:23 jcdutton: Ah! I just spotted, its not liking the IRQ. IRQ11 should be the gpu one
21:23 karolherbst: I was confused
21:23 karolherbst: nvm
21:23 karolherbst: ohhh
21:23 karolherbst: yeah, that might explain it
21:24 karolherbst: but.. I think we are running into this issue somewher else as well
21:24 karolherbst: jcdutton: can you check if fixing up the IRQ helps?
21:25 jcdutton: This is the first time I have tried passthru. What do I need to try to fix the irq ?
21:26 jcdutton: It looks like the irq fires before the nouveau driver probes.
21:37 jcdutton: I think I maybe need to look into how to get kvm to passthru msi interrupts