#
✕ nick:marcan
#asahi · 2022-09-11 09:35
marcanno idea, but here's a guess: it's running something every frame, and our simple framebuffer has no vsync so that runs uncapped
#asahi · 2022-09-10 18:33
marcanI know, and that's why I'm telling you, I hope you can learn from it
#asahi · 2022-09-10 18:29
marcanjust leave these things to us in the future, please (until something actually happens and is officially announced)
#asahi · 2022-09-10 18:28
marcan(and because people don't learn if you don't say anything)
#asahi · 2022-09-10 18:27
marcanbecause I don't appreciate people speaking on behalf of the project if they can't do it with tact and care, it hurts everyone else including the other developers
#asahi · 2022-09-10 18:26
marcan(or what might happen in the future)
#asahi · 2022-09-10 18:25
marcanit's not your job, you're not involved in the decision, and you didn't even do a good job of expressing what actually happened
#asahi · 2022-09-10 18:25
marcanmps: please don't poorly re-tell project decisions that haven't even been fully decided yet like that to random new users
#asahi · 2022-09-09 09:52
marcanI don't think we compile apparmor into the default kernel, no
#asahi · 2022-09-07 05:23
marcanother than OBS being slow in general because it relies on the GPU, so of course it's going to be extremely slow with CPU rendering
#asahi · 2022-09-07 05:23
marcanyou'd probably have the same result on an intel machine with the same camera
#asahi · 2022-09-07 05:22
marcanit's just different UVC drivers with different buffering. nothing to do with asahi most likely.
#asahi · 2022-09-06 15:37
marcanI don't know if our RCs support it, would have to check
#asahi · 2022-09-06 15:32
marcan10 bit tags?
#asahi · 2022-09-06 09:05
marcanif you regard "popping" as potentially hazardous, you should never use bluetooth audio, since they can always happen :)
#asahi · 2022-09-06 09:05
marcan"popping" is just skipping due to packet loss, and is a normal thing on bluetooth audio (though it's worse here because btcoex isn't ready)
#asahi · 2022-09-05 04:54
marcanstreaming some wifi stuff in a minute, #asahi-stream as usual
#asahi · 2022-09-04 11:24
marcanand ~zero throttling
#asahi · 2022-09-04 11:24
marcanI bet that's a nice mix of the CPU just being better, and also much lower wake latencies
#asahi · 2022-09-04 11:23
marcangood to hear it's way better than Intel with USB hardware though - that's what I was looking forward to
#asahi · 2022-09-04 11:18
marcanproaudio is hard enough when you don't throw in bluetooth into the mix ;)
#asahi · 2022-09-04 11:17
marcanthis stack is too deep, it could be anything
#asahi · 2022-09-04 11:17
marcanthe bluetooth host controller hardware could be involved too, potentially
#asahi · 2022-09-04 11:16
marcane.g. it reports playback timestamps in the past, and ardour thinks it's already late for the next cycle when it isn't
#asahi · 2022-09-04 11:16
marcancould also just be that some kind of timestamping involved in the audio backend is broken for bluetooth
#asahi · 2022-09-04 11:14
marcanyou'd have to give details of exactly how you have your system configured (JACK/PA/etc and what Ardour backend) to try to reproduce it, but it might only happen with certain Bluetooth devices too since the codec plays a role
#asahi · 2022-09-04 11:14
marcanbluetooth output isn't really intended for proaudio, and I doubt anyone really optimized/debugged that kind of audio path for this use case
#asahi · 2022-09-04 11:13
marcanit could be you ended up with a ridiculously small buffer size causing actual high CPU usage, or that whatever backend contraption you're ending up with is not actually giving Ardour reasonable real-time processing time for whatever reason
#asahi · 2022-09-04 11:12
marcanwith something like bluetooth with high latency, who knows how that is being calculated
#asahi · 2022-09-04 11:11
marcanDSP usage is relative to how much time there is in a processing cycle to complete DSP tasks on time
#asahi · 2022-09-04 11:11
marcanCy8aer[m]: "DSP usage" is not a CPU usage meter (check your actual CPU usage)
#asahi · 2022-09-02 05:45
marcanno, it's new codecs, it's not even HDA
#asahi · 2022-09-01 06:48
marcanManouchehri: does it still happen after `pacman -Syu`?
#asahi · 2022-08-31 18:30
marcanI have a pending thing to test
#asahi · 2022-08-31 18:30
marcanit'd possible the alloc stuff is still broken, just not as badly
#asahi · 2022-08-31 18:12
marcancheers
#asahi · 2022-08-31 18:11
marcan(not an official policy or anything, most people just tend to default to that I find)
#asahi · 2022-08-31 18:10
marcan(we're pretty loose with bug/mechanical/etc fixups, as long as the person submitting them consents to squashing without attribution)
#asahi · 2022-08-31 18:08
marcanalso you might want to move this stuff to #asahi-dev
#asahi · 2022-08-31 18:07
marcan:)
#asahi · 2022-08-31 18:06
marcanso getting the latter upstreamed will certainly help :)
#asahi · 2022-08-31 18:06
marcan(but it won't touch the underlying smc gpio thing)
#asahi · 2022-08-31 18:06
marcanwhich is coming up soon
#asahi · 2022-08-31 18:06
marcandon't try to upstream the actual pcie enable gpio thing though - *that* needs a rework
#asahi · 2022-08-31 18:01
marcanyeah, rtc can go in later (spmi needs a rework anyway)
#asahi · 2022-08-31 18:00
marcanyou probably need to upstream the mailbox/rtkit atomic support too, it's a dep for some of the SMC code
#asahi · 2022-08-31 17:59
marcanrmk: nope, assuming you keep commit ownership etc the usual way, go for it :)
#asahi · 2022-08-31 17:35
marcanrmk: not really, it's just on the back burner
#asahi · 2022-08-31 17:34
marcanI did notice the latest version ends up with yellow text, not sure if that's intentional, but whatever :-)
#asahi · 2022-08-31 17:34
marcangrub is fine on asahi. there was a less critical failure (falls back to text mode) but we got it fixed pretty quickly.

plus ›