00:49karolherbst: imirkin: mhh.. so if I'd like to debug an issue on a "nv68" the nv40 (or nv44?) GPUs are probably the closest one to it, right?
00:50imirkin: i forget if they tend to be the nv44 or nv40 graph class
00:50imirkin: (there's no logic to which GPUs have which one)
00:50karolherbst: well.. doubtful it matters if apparently the entire system crashes/freezes (https://gitlab.freedesktop.org/drm/nouveau/-/issues/123)
00:51imirkin: yeah, i saw that issue
00:51imirkin: really nothing i can say that's helpful on there though
00:51karolherbst: I just want to see if I can trigger it on my system and if not... not much I can do to help anyway
00:52imirkin: it sounded like some issue triggered by the default compositor
00:52karolherbst: yeah... not sure
00:52karolherbst: lxde didn't really had a compositor back then
00:53imirkin: yeah, i dunno enough of the specifics
00:53karolherbst: probably some weirdo ddx bug
00:53imirkin: nah, ddx is pretty solid
00:53imirkin: unless the modesetting one is being used
00:53imirkin: yeah, dunno
00:53imirkin: should double-check what the hack does
00:54karolherbst: checks for nv50+
00:54imirkin: whether it accounts for nv68 being pre-nv50
00:54imirkin: it might, it might not. dunno how it works
00:54imirkin: a common way to check is chipset == 50 || >= 80
00:54karolherbst: let's see...
00:54karolherbst: if (args == 0x050 || args >= 0x80)
00:55imirkin: it's almost like the person who wrote it was familiar with nvidia chipset id's
00:56karolherbst: ohh btw.. that weirdo vmm race is fixed, so we should start seeing way less issues with the modesetting ddx (hopefully)
00:56imirkin: i'm still on v5.6
00:56imirkin: afraid to upgrade, so many issues crept into like 5.8-5.10
00:56karolherbst: :( yeah...
00:57karolherbst: I think 5.15+ should be better though
00:57karolherbst: or any stable one :D
00:57imirkin: let's hope!
00:57karolherbst: 5.14 even I think
00:57karolherbst: still scanning through the Mailing list, but I ddi picked up a few fixes
00:58karolherbst: at least on my desktop where I run nouveau I didn't encounter any serious issues anymore
01:00imirkin: it's frightening that trees were getting pushed where you WERE seeing serious issues :p
01:00karolherbst: imirkin: btw.. if you still have fixes which aren't merged... I can push those if you want to :D
01:00imirkin: i dunno, looks like you have that vzalloc patch from like ... a year ago
01:00karolherbst: ohh, right
01:00imirkin: i'll check if i had anything else pending
16:11karolherbst: skeggsb, imirkin, Lyude (although I know you already reviewed it): I would like to get some fixes in asap (internal CVE bug and shit, easier to just apply the fixes than saying "ehh... "), so if you want to take a look before I merge it, please do :D https://gitlab.freedesktop.org/drm/nouveau/-/merge_requests/14
16:12imirkin: i'll look at it in the next 30 mins. after that, no news is good news.
16:16imirkin: man, gitlab really stinks for reviewing stuff. sigh.
16:21imirkin: unfortunately i don't have much comment on these lifetime issues
16:22imirkin: i'm not up to date on how the drm stuff works either, and while i could read up on it, so could anyone else
17:52Lyude: karolherbst: will go through and address your review comments at some point today
19:06imirkin: karolherbst: a ton of people have reported issues with fermi GPUs and CE firwmare init
19:06karolherbst: I also think I saw something but...
19:06imirkin: another of the many reasons i haven't updated
19:07imirkin: 02:00.0 VGA compatible controller : NVIDIA Corporation GF108GL [Quadro 600] [10de:0df8] (rev a1)
19:07karolherbst: so if it's a regression you could bisect it :D
19:07karolherbst: I think I already know what broke it, but...
19:07imirkin: yeah. or i could wait for it to be fixed.
19:07imirkin: this is my primary desktop
19:07karolherbst: wouldn't be surprised if it's the same thing which happened on tesla
19:07imirkin: with WFH, making sure it runs has become more important than usual.
19:08imirkin: amongst many others, yea
19:09imirkin: you can see my theory, that was just from glancing at some code + bisected commit
19:09karolherbst: right.. but usually you can still use the older kernel untless your firmware sucks or you don't use a bootloader :p
19:09karolherbst: without looking at code
19:10karolherbst: we need more testing :D
19:10imirkin: yeah, i can revert kernels. when the thing totally doesn't work, that's great
19:10imirkin: but when it crashes middle of the day - not so great
19:10karolherbst: imirkin: what I pushed last (or so)week: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/nouveau?h=v5.15&id=d1d94b0129dccd226784633c60b7df90e8a051b5
19:10imirkin: if push comes to shove, i can do some testing, but i'd rather let people with dedicated hw and time take care of it :)
19:11imirkin: that looks like an unrelated bit.
19:11karolherbst: so I wouldn't be surprised if the arch after tesla is also weird and broke due to the engine_id stuff
19:11imirkin: oh. yes. perhaps.
19:12karolherbst: especially as it was cc: stable 5.12+ :D
19:12imirkin: but the patch you mention won't fix this, unless it's somehow very circuitously related
19:12karolherbst: yeah.. it's just for tesla
19:12karolherbst: just saying that fermi probably has a similiar bug
19:12karolherbst: I _think_ I have a fermi GPU somewhere... let's see
19:12imirkin: do you have time + hw to play with it?
19:13karolherbst: okay.. I don't have a fermi afterall
19:17karolherbst: mupuf: how much time do you have in regards to ship me a few fermi boards?
19:20karolherbst: I have like everything else.. maybe something prior nv40 would also help as long as it's PCIe
19:20karolherbst: but fermi is what I'd need I think
19:31imirkin: there are some nv3x's running around on PCIe bridges
19:31imirkin: i have one :)
19:31imirkin: and even a nv17 or nv18
19:31imirkin: i forget if i have that one or not
19:39karolherbst: ahh cool
19:39karolherbst: that would be useful to get
19:39karolherbst: having only PCIe slots and such
19:40karolherbst: if there are TB to PCI/AGP bridges? :D
19:40karolherbst: but not sure I already have TB.. probably just USB-C
20:10mupuf: karolherbst: well, i\m sure I can find some time to go through my GPUs and pack them
20:12mupuf: karolherbst: what chipsets do you have already?
20:13mupuf: I have some pretty beefy stuff I can send, including a dual nvc0 and a titan
20:29karolherbst: mupuf: I usually have everything starting with Kepler, and some nv44
20:29karolherbst: three teslas, but not sure what
20:30imirkin: mupuf: if you really karolherbst, send him a G80
20:30imirkin: ... really *hate* ...
20:30karolherbst: imirkin: do you know if I had a nv44 or nv40 class nv4x?
20:30karolherbst: the one with native support for.... this thing
20:30imirkin: i think the nv40 one
20:30imirkin: coz you were able to fake the nv44 one
20:30imirkin: and iirc nv40 is the one that supports index buffers
20:30imirkin: or something
20:31karolherbst: it was nv44
20:31karolherbst: nv40 :D
20:31imirkin: good typo's to make.
20:32mupuf: imirkin: ha ha, yeah, the nv50 is a weird beast!
20:32karolherbst: mupuf: I can make a complete list though
20:32imirkin: the G80 just has a lot of ... annoying restrictions
20:32mupuf: I guess I should make the list, and see what you want
20:32karolherbst: should only take... 20 reboots
20:32imirkin: also among other things, its GS varyings map works differently. because it can.
20:32mupuf: imirkin: power management was also different
20:32karolherbst: ohh wait.. I have three ports
20:32imirkin: yeah, not surprising
20:32imirkin: it also has that weird memory layout
20:33imirkin: 320MB -- bigtime memory :)
20:35mupuf: that's smaller than the rumoured cache of the RDNA3 7900 XT :D
20:44karolherbst: oh nice....
20:44karolherbst: on the one nv42 my machine doesn't boot...
20:44karolherbst: hangs at parsing the vbios I think...
20:46karolherbst: ohh wait..
20:46karolherbst: it was this "GPU not primare, system just messes up" bug
21:01Lyude: karolherbst: you haven't made any changes to https://gitlab.freedesktop.org/drm/nouveau/-/merge_requests/14/diffs?commit_id=7669390cf7e84ff72679e021ea981437c7eec6c1 since I last reviewed it for jcline did you?
21:02karolherbst: I didn't touch any of those commits unless for rebases + tags
21:02karolherbst: and adding that simple note
21:04imirkin: karolherbst: FX 3450?
21:04Lyude: karolherbst: reviewed, will try to get to the review comments you left on the CRC work tomorrow
21:04imirkin: that one only works if it's the only GPU in my system, in the first PCIe slot. also remember it has an extra power connector.
21:06karolherbst: imirkin: some NV42
21:06imirkin: yeah, that's the one i have :)
21:06karolherbst: machine stopped doing anything after "[ 13.573935] nouveau 0000:02:00.0: bios: version 05.41.02.43.05"
21:07karolherbst: not even the reboot button worked :D
21:07imirkin: oh, for me it didn't even POST
21:07karolherbst: oh wow
21:08karolherbst: although that's one a UEFI system here
21:08imirkin: ah yeah, no UEFI here
21:08imirkin: plain ol' BIOS
21:08imirkin: i don't have one of them fancy new boxes...
21:08karolherbst: ehh "[ 11.796132] nouveau 0000:01:00.0: gpio: init failed, -22"
21:08karolherbst: "[ 11.783260] nouveau 0000:01:00.0: gpio: GPU is missing power, check its power cables. Boot with nouveau.config=NvPowerChecks=0 to disable." :D
21:09imirkin: well, more likely that's because gpio failed
21:09karolherbst: no. I didn't connect the power :D
21:09imirkin: oh. well there's that.
21:09karolherbst: glad to see that the code actually works :D
21:10imirkin: yeah, didn't your intern add that?
21:10imirkin: someone did
21:10karolherbst: yeah, I think so
21:10karolherbst: now it doesn't fail, waht a surprise
21:12karolherbst: GPUs are getting bigger and bigger over time...
21:12karolherbst: my ampere one is more wide than the actual mobo
21:13imirkin: more *wide*?
21:13karolherbst: longer I mean
21:14imirkin: "full length" pci cards? :)
21:15imirkin: they clip in the back too
21:15karolherbst: mhh maybe
21:15karolherbst: I think it's longer though
21:16karolherbst: ahh no..
21:16imirkin: full length is pretty long.
21:16karolherbst: yeah.. 42mm longer
21:16karolherbst: but my case has the advantage that I don't have this weirdo metal stuff for the dvd player + HDDs
21:16karolherbst: so plenty of space
21:17karolherbst: there are holes to add it though if you really want to
21:17imirkin: and remember that e.g. dvd drives are "half height" 5.25" devices
21:17imirkin: the corollary being that there were "full height" devices too
21:17imirkin: (hard drives)
21:17imirkin: which would take up 2x modern 5.25" bays
21:17imirkin: for some huge 80MB of storage or something
21:17karolherbst: yeah.. but my PSU and the space for HDDs is at the bottom :)
21:18imirkin: yeah, in my case too
21:19imirkin: i started out with 5x 3.5" drives. now down to 2, since we live in the future.
21:19karolherbst: yeah.. I have 2 nvme SSDs :D
21:19imirkin: might soon be getting time to upgrade those drives agai
21:19karolherbst: nvme is really worth it
21:19imirkin: every 5y sounds wise?
21:20karolherbst: I'd rather decide based on new technology
21:20karolherbst: the jump from SATA SSDs to nvme ones is huge btw
21:20imirkin: these are spinnig rust, so ...
21:21imirkin: 2x 8TB
21:21imirkin: maybe i can get 2x 16TB?
21:21karolherbst: not worth it really
21:21imirkin: how do you mean?
21:21karolherbst: to invest in spinning rust :D
21:22imirkin: i mean ... how else do i get the storage?
21:22karolherbst: you probably need this amount of space though I guess?
21:22imirkin: need is a relative concept
21:22imirkin: but i like it :)
21:23imirkin: started out with 5x 2TB in a RAID-5 in 2010 or so
21:23karolherbst: there are 16 TB SSDs
21:23karolherbst: though they cost a bit
21:23imirkin: then moved to 2x 8TB in a RAID-1
21:23karolherbst: 8TB seems reasonable
21:23karolherbst: just one grand
21:24karolherbst: or go with multiple 4 TBs ones for ~400 each
21:24imirkin: hm, like $300 for a 16TB HDD
21:24karolherbst: anyway.. 1TB/2TB are the sweet spot for storage/prices for SSDs
21:24imirkin: (* 2 for the RAID)
21:24karolherbst: yeah.. but HDD is slow
21:24imirkin: what it lacks in speed, it makes up for in size :)
21:24karolherbst: and noisy
21:25imirkin: i'd rather have 2x 16 TB SSD's, yes.
21:26karolherbst: I am actually thinking of doing some external storage system passively cooled with like a lot of SSDs :D
21:27karolherbst: there are cases where you can put like 12 into it
21:27imirkin: fwiw my "system" drive is a SSD
21:27imirkin: the spinning rust is just for my hoarding ways
21:28karolherbst: apparently it can cool up to 105W passively for the CPU + 65W for the GPU ....
21:29karolherbst: sadly if you buy this 240W PSU it takes away a bit of space, so no 12x 2.5" drives anymore :D
21:29karolherbst: but anyway.. thinking about something like that
21:30imirkin: that's pretty neat
21:32karolherbst: mupuf: that's what I have: https://gist.githubusercontent.com/karolherbst/41a6613eaafd8ec0e3d53f408aa9bc37/raw/6c1813ba9ddba810911371219654ef999b204846/gistfile1.txt
21:32imirkin: all plugged into one thunderbolt connector ;)
21:33karolherbst: .... soo.. I know you are making jokes, but this mobo apparently has this "mining" stuff, where you can plug like a lot of GPUs into it with a special cable or whatever
21:33imirkin: soon your office will look like jrayhawk's
21:34karolherbst: I found a perfect shelf for the GPUs though, so it's not that bad
21:34imirkin: this is the money shot: http://heybryan.org/shots/setups/jrayhawk/20120512_008.jpg
21:35imirkin: i esp like the video cards sorted by height of VGA connector on the top shelf
21:36karolherbst: oh wow :D
21:36soreau: you know you've gone to far when you have computer cases hanging from the ceiling
21:36karolherbst: if GPUs wouldn't be all diferent in height these days, it isn't actually a bad idea :D
21:37karolherbst: soreau: actually, that's not what concerns me, it's rather that this person trusts the cable plugs to hold the HDDs :D
21:37soreau: karolherbst: hahaha
21:37karolherbst: also.. the keyboard and the mice? it's not ergonomic to sit/work like this :P
21:38imirkin: that's the take-away from that photo ... "it's not an ergonomic setup"
21:38imirkin: it's perfect, just give the man a split keyboard
21:38karolherbst: well.. the desk also needs to be a little bigger :D
21:38karolherbst: I think I have more desk space, but less on it :D
21:39imirkin: yes, well i suspect that space is a finite resource here.
21:39soreau: karolherbst: oh actually, if you look more closely, there are zip-ties around the sata connectors on the hdd's
21:39karolherbst: that's makes it better
21:39soreau: cable management at it's finest lol
21:40karolherbst: I like the chain
21:40jrayhawk: It's an okay setup, but it could use some more monitors.
21:41karolherbst: jrayhawk: how do you even :D
21:42karolherbst: although I am usually going the other way and have a very minimalistic setup
21:43jrayhawk: karolherbst: your advice from 2020-02-05 to use wayland turned out to be 100% accurate
21:43jrayhawk: I was amazed when it just worked.
21:43karolherbst: but honestly.. what disturbs me most is how uncomfortable it looks like to sit there in regards to the keyboard and mice.. I could live with everything else :D
21:43karolherbst: jrayhawk: yay
21:43jrayhawk: I gave up on sitting.
21:44jrayhawk: It's been a standing desk for the past 6 years.
21:44karolherbst: I actually have a motorized desk
21:44karolherbst: it's really cool to stand once in a while
21:44soreau: everyone needs a walk-desk, like Linus
21:45soreau: basically treadmill+desk
21:46karolherbst: yeah, I saw it from this one documentary once
21:47soreau: so when is nvidia releasing for-foss fw? or will it ever happen
21:48karolherbst: I doubt this will ever happen as this kind of bites itself with the idea behind signed firmware
21:49karolherbst: _maybe_ there will be source at some distant point in future, but how good is it, as long as you can't run any modded firmware anyway
21:50soreau: huh, I thought there was some promise
21:50karolherbst: was there?
21:51soreau: prolly not
21:51soreau: oh well, at least gbm works on the prop driver now
21:51karolherbst: yeah, it does
21:51imirkin: karolherbst: insinuations of firmware being released to make nouveau more usable
21:51karolherbst: kind of
21:51karolherbst: imirkin: I guess so
21:52karolherbst: jrayhawk: btw, using gnome or something else?
21:52soreau: (wayfire ftw)
21:52karolherbst: never tried that
21:53karolherbst: I kind of gave up on the idea using "lightweigt" desktops because they tend to draw more power than full blown ones like gnome or plasma
21:54karolherbst: although I think it got better as people started to move sane defaults into the kernel and not making userspace responsible for setting better settings
21:55karolherbst: Lyude: I think we need to talk about what "approved this MR" should translate to inside an MR
21:56karolherbst: as now I moved to using dim for scripting away pain and not marge bot, I might even make use of it inside dim for auto tagging
21:56soreau: karolherbst: well wayfire is worth the whirl if not just for the eye candy shaders
21:57soreau: blur plugin is the only thing kinda heavy in wayfire and I think it's off by default
21:57karolherbst: well.. the problematic things are not the "heavy" things, but the time and dedicated to get stuff power efficient and to deal with power hungry apps/daemons
21:58karolherbst: just because something looks more fancy, doesn't mean it draws more power, and stuff looking plain isn't more power efficient automatically
22:05jrayhawk: An eye-candy fan, eh? Maybe you would know... I'm looking for a way to straight black-to-alpha composite some X11 windows on top of eachother, and I can't find any compiz plugins that actually do this. Is there a pre-existing solution for such a thing, or am I stuck writing my own compiz plugin?
22:06imirkin: black-to-alpha composite? i.e. black = transparent?
22:07soreau: jrayhawk: I wrote this for wayfire
22:07imirkin: don't think any X11 compositors support color keys offhand
22:07imirkin: or you're on wayland i guess?
22:07soreau: it's called keycolor https://github.com/WayfireWM/wayfire-plugins-extra/blob/master/src/keycolor.cpp
22:07imirkin: so this is Xwayland windows composited by the wayland compositor?
22:07jrayhawk: I would prefer X11 in this case, but I can probably work out how to suffer wayland in this case.
22:07soreau: jrayhawk: wayfire is the closest thing to compiz on wayland
22:08imirkin: jrayhawk: x11compmgr will let you set an alpha on a whole window, but i don't think it supports color keys out of the box
22:08jrayhawk: soreau: Okay, thanks, I will give that a shot.
22:08soreau: jrayhawk: yea after it's all setup, you just give it a color and threshold and it turns that into alpha with a shader
22:08soreau: on all surfaces
22:11soreau: jrayhawk: for compiz the closest thing that comes to mind is the colorfilter plugin but you pass that an ARB fragment program file and it runs the instructions but I don't think it can do alpha? you'd have to tinker with it to see
22:14jrayhawk: I am hesitant about Wayland only because screen capturing I was hoping to do on this has instructions that are all "use git development HEADs of this stack of four different utilities and pass three environment variables and it might work if you have the right session management and driver"
22:14jrayhawk: which is the sort of thing that will probably resolve itself in 3 months.
22:15imirkin: jrayhawk: should be achievable to hack xcompmgr for this
22:15karolherbst: I guess this issue is way too compositor specific in the end
22:16soreau: jrayhawk: wf-recorder is a nice program fro screen capture
22:16soreau: for wlroots compositors*
22:17karolherbst: I really hope we get common extensions for this kind of basic stuff soonish :p
22:17soreau: wlroots just landed in fd.o namespace/server/gitlab so seems like good news
22:18karolherbst: soreau: what do you mean precisely here?
22:18soreau: karolherbst: I mean that wlroots moves from github to gitlab.freedesktop.org
22:19soreau: and this happened like today
22:19karolherbst: I meant more that I don't think we have a stable "common" screen recording API yet
22:20soreau: on that front I'm hoping that more wayland protocols are merged to accomodate such things
22:20soreau: wlr-protocols was also moved
22:20karolherbst: I don't know what gnome even implements
22:20soreau: mutter :P
22:21karolherbst: yeah.. I meant for screen recording
22:21soreau: I'm sure they have some in-house solution :P
22:21Lyude: there is
22:21Lyude: it's pipewire :P
22:21karolherbst: Lyude: you still need an API for that though
22:22soreau: yea pipewire, that's the thing that if you don't have it, gtk apps take forever to start u pin wayland :P
22:22soreau: up in*
22:22Lyude: also tbh, if you don't have it: why lol
22:22karolherbst: yeah.. pipewire is soo much better than pulse
22:22karolherbst: ditch pulse, become one of the cool people
22:22Lyude: it is technically currently the best sound system on linux, being that it's the only one that can actually keep things perfectly in sync along with not suffering from pulse's latency issues
22:23soreau: ok I'm easy, you've convinced me
22:23karolherbst: I've heard great things from pro audio people actually about pipewire
22:23karolherbst: because you don't need to deal with jack and pulse anymore :D
22:23soreau: but it supports jack and pulse apps?
22:23imirkin: i've been using pipewire to deal with BT headset
22:23imirkin: seems largely successful
22:23karolherbst: for jack you usually need to install some pipewire-jack package though
22:24karolherbst: pulse usually is supported ootb
22:24imirkin: pulse also supports all this, but i'm pretty allergic to pulse for unclear reasons
22:24imirkin: seems like it's very heavy
22:24Lyude: Same, I use pipewire's bluetooth LDAC support for my WH-1000XM4
22:24karolherbst: what's cool about pipewire is, that you can just use the jack apps to rewire your audio channels :D
22:24karolherbst: on the fly
22:24imirkin: there aren't native controlling utilities, so i still have the pulse compatibility layer
22:24imirkin: so i can use pavucontrol
22:24karolherbst: it's like pavucontrol on steroids
22:24karolherbst: imirkin: pavucontrol is shit
22:24imirkin: all i need to do is switch profiles
22:25imirkin: on the headset
22:25imirkin: it has a dropdown
22:25karolherbst: ahh yeah..
22:25imirkin: the end.
22:25karolherbst: don't need pavucontrol for it usually
22:25soreau: I live behind the times though, buy older hw after all reviews are out, install new software after I decide it's actually better
22:25imirkin: doing it directly in pipewire cli is pretty painful
22:25karolherbst: imirkin: I meant a normal audio widget usually allows this as well
22:25karolherbst: but why do you need to switch profil... ohhh
22:25imirkin: pavucontrol is that widget for me ;)
22:25imirkin: coz BT sucks
22:25Lyude: you need to switch profiles to change the bluetooth codec
22:25karolherbst: because the headset thing is crappy compared to the headphones one
22:26imirkin: either you can do microphone
22:26karolherbst: I see...
22:26imirkin: or you have stereo audio
22:26karolherbst: I remember
22:26soreau: imirkin: does it always pick the high fi profile without the mic or something?
22:26karolherbst: I think it got fixed with this new fancy thing
22:26imirkin: soreau: that's configurable
22:26imirkin: but most of the time i want to listen to music
22:26imirkin: which sounds much better when it's not reduced to 8khz quality
22:26Lyude: yeah for calls i usually still use my laptop's mic anyway
22:26soreau: imirkin: for me, it always defaulted to stereo high quality, without mic
22:26Lyude: since it's one of those nice intel DSP ones
22:26imirkin: ah heh. yeah, that'd work. no mic here (nor laptop)
22:27karolherbst: imirkin: isn't that better with A2DP?
22:27imirkin: karolherbst: A2DP is unrelated to this
22:27imirkin: A2DP allows you to send like next/pause/play commands
22:27karolherbst: but it supports also various codecs
22:27imirkin: this makes audio sound better
22:27karolherbst: and input + output
22:28Lyude: yeah this mostly has to do with mSBC being the bluetooth audio codec with mc support
22:28imirkin: if you want input + output, you're back to headset profile
22:28Lyude: what pipewire -should- be doing is switching between LDAC and mSBC on the fly when the microphone isn't/is in use
22:28Lyude: (which is what android does)
22:28imirkin: that'd be neat.
22:28karolherbst: Lyude: so you speak and it switches?
22:28karolherbst: or depending on what applications use?
22:28imirkin: karolherbst: no, you launch an app that wants to record and it switches
22:28Lyude: more based on like, if an app is reading mic input
22:29karolherbst: but I was under the impression there was a fix for that...
22:36karolherbst: why is it 2021 and I need to restart apps after switching profiles... :(
22:37imirkin: i definitely don't need to with pipewire
22:37imirkin: i have it set as the default alsa backend
22:37Lyude: karolherbst: it's just for turning the mic on/off, also ldac support is definitely very new
22:37imirkin: if i launch an app without pipewire running, and then i start pipewire, then that app's audio needs to be reloaded
22:37karolherbst: might be that those apps are also just broken
22:38imirkin: anyways, this is all hardly perfect
22:38imirkin: the audio often crackles/etc
22:38Lyude: i haven't had any audio crackling actually
22:38imirkin: yeah, dunno if my comp is slow or my hardware is shit
22:38karolherbst: seems like mSMC is better than CSVD but still crappy
22:38imirkin: like the btusb adapter/etc
22:38imirkin: iirc i need to update kernel for mSBC to work
22:39karolherbst: but my audio setup is at least 4 times as good as yours, because I have 192k and you don't :p
22:40imirkin: sadly, this all worked much better bakc when bt audio was in the kernel
22:40imirkin: i remember headset working fine back in like ... 2003
22:40Lyude: karolherbst: I didn't think the human ear went that high? lol
22:40Lyude: 96 KHz is already pushing it
22:42karolherbst: Lyude: yeah.. not sure.. the audio stuff wasn't more expensive than a normal 48k one, so...
22:42karolherbst: just wanted some proper audio device and not suffer from crappy on board audio noise