ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
heat has joined #dri-devel
hanetzer1 has joined #dri-devel
Daanct12 has joined #dri-devel
hanetzer has quit [Ping timeout: 480 seconds]
Daanct12 has quit [Ping timeout: 480 seconds]
camus1 has joined #dri-devel
camus has quit [Ping timeout: 480 seconds]
GloriousEggroll has quit [Quit: Death is life's way of telling you you've been fired.]
Daanct12 has joined #dri-devel
Daanct12 has quit [Read error: Connection reset by peer]
Lightkey has quit [Ping timeout: 480 seconds]
ngcortes has quit [Remote host closed the connection]
Lightkey has joined #dri-devel
Daanct12 has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
Guest24 has quit [Remote host closed the connection]
camus has joined #dri-devel
camus1 has quit [Read error: Connection reset by peer]
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
boistordu has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
boistordu_ex has quit [Ping timeout: 480 seconds]
lemonzest has quit [Quit: Quitting]
<zmike> I gotta say it: ci today is A++ would assign to marge again
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
<zmike> obviously the second I say that, the stoney jobs are revealed to have been cast as the bad guy in an 80s action movie which is just now reaching its climax
Daaanct12 has joined #dri-devel
Daaanct12 has quit [Remote host closed the connection]
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
khfeng has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
<agd5f> vsyrjala, We have something similar (vram self refresh on dGPUs), but it's a trade off, more power vs. the overhead of evicting vram on suspend. That said, with the current PRIME use cases, there isn't much that you actually care about in vram once the frame is rendered so we just turn off the memory. May have to revisit that at some point.
<agd5f> mdnavare, two unsolved problems with runtime pm for dGPUs: 1. is there is a console bound to the dGPU, you can't put it to sleep since there is a kmap of the framebuffer BAR for the console. I guess we need some kind of fault handling thing to handle that properly when someone accesses it. 2. display hotplug when a dGPU is powered off and you don't get HPD interrupts. Not sure how to solve this efficiently in the kernel. Probably need support
<agd5f> from userspace to query the GPUs on some user input.
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
<airlied> agd5f: some platforms have acpi workarounds for hotplug irqs
<airlied> though I'm not sure if those are OEM hacks or std
<agd5f> airlied, yeah, for laptops, but nothing like that for add-in boards
<airlied> nope indeed not, for desktops that does suck alright
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
<airlied> karolherbst, jenatali : clang CL 3.0 fun aside https://lists.llvm.org/pipermail/cfe-dev/2021-July/068522.html
<vsyrjala> pme should be a thing no?
<vsyrjala> or is that one of those "we didn't bother implementing it properly" things?
<agd5f> vsyrjala, sure, if the hw supports it.
<agd5f> not worth the extra always on block
<jenatali> airlied: I thought there were command line options to enable/disable them
<jenatali> Been a little while since I looked though and I've paged it all out
<vsyrjala> our dgpus should support it. so assuming the mobos don't suck it should work. which makes our dgpus nicer than our igpus actually
<airlied> jenatali: have to dig through the new ext mechanism to work out what the cmd lines opts might be
<jenatali> airlied: ping me tomorrow and I can probably dig it up if you haven't found it by then
<jenatali> I think it was -cl-ext=-<something> to turn it off
Daanct12 has quit [Remote host closed the connection]
<airlied> jenatali: ah cool will rip that off then
<airlied> jenatali: clover was relying on clover just doing the C defines
<jenatali> airlied: yeah I thought I could just manipulate the features and get it working but the headers weren't ready yet
<jenatali> If things are better now that'd be great
<airlied> of course we don't pass cmd line args like that, so I'll have to dig more
<airlied> jenatali: yeah I'm trying to fix the headers
<airlied> but got stuck because clover doesn't work anymore
<jenatali> Sweet
Peste_Bubonica has quit []
<jenatali> airlied: I'd suggest switching to pass command line args like that
<airlied> jenatali: we don't use the driver at all in clover
<airlied> my previous attempt to fix that uncovered lots of other pain
<jenatali> That seems like a mistake IMO
<jenatali> I still have a backlog item to try to converge CLOn12's clang invocation with clover
<airlied> there was a bunch of tradeoffs in which way you go
<jenatali> I hadn't seen any cons this way so far
<airlied> yeah I suspect it's got something to do with the amd native backend
<jenatali> Ohhh yeah that could do it
<airlied> was where I started converting it
heat has quit [Ping timeout: 480 seconds]
<airlied> yay vector_swizzle doesn't die
<jenatali> airlied: let me know there's something I can help with, convergence here would be great
<airlied> jenatali: do you still use opencl-c.h?
<airlied> jenatali: that is another project to move away from with newer clang
<jenatali> airlied: yeah
<jenatali> Are they switching to the libclc headers?
<airlied> nope they are generating the builtin defines on the fly
<airlied> from tablegen
<jenatali> That'd be a pain, we embed the headers, and going from 2 to 30 is going to be annoying
<airlied> or something nuts like that
<jenatali> Oh builtins are better
<jenatali> Yeah I remember seeing that discussion now that you mention it
<jenatali> They wanted to ensure 0-regression before switching or something
<airlied> yeah I'll stick with fixing up opencl-c.h first I suppose then at least can test for switching
<airlied> https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11752 seems to let me move forward for now
<jenatali> Can't you just pass -cl-std=cl1.2 on the command line (or the equivalent) too?
<jenatali> I thought that also disabled all the optional stuff
<airlied> jenatali: I want cl 3.0
<airlied> at least I want the overrides to work
<jenatali> airlied: oh, right
<airlied> ughh phab only lets you submit diffs one at a time, guess that's how it going to be then
<jenatali> airlied: just wait til they finish switching fully to github
<zmike> airlied: you could use git-phab?
<airlied> zmike: does it do some magic? they have some tool called arc I should look at
<zmike> yes, it lets you put up a series in a single differential
<zmike> though if you're not going to be using this workflow a lot, it might not be worth the hassle of setting up
<zmike> you need arc too
* airlied can handle email or gitlab, all these half way houses need to go away :-P
<zmike> amen
<airlied> zmike: cool will look a bit at it, but might just go the lazy brain way and create lots of differentials
<zmike> I PMed you some info, may or may not be worth using
<zmike> yea
<airlied> oh woah the atomic funcs are all kinds of messed up, and my previous patch to reorder them doesn't apply clean
<airlied> everytime I work on llvm, I feel like playing Johnny Cash Hurt
<jenatali> airlied: that's why I stopped cl3.0 attempts when I realized llvm wasn't ready
yk has quit [Remote host closed the connection]
tzimmermann has joined #dri-devel
mattst88 has quit [Remote host closed the connection]
mattst88 has joined #dri-devel
mattst88 has quit []
mattst88 has joined #dri-devel
mattst88 has quit []
mattst88 has joined #dri-devel
aravind has joined #dri-devel
<mdnavare> airlied: Is there a sysfs that can tell what GPU is it using for rendering?
sdutt has quit [Remote host closed the connection]
<mdnavare> agd5f: vsyrjala: How does Nvidia handle these issues regarding the console BAR and hotplugs? I was reading this: http://download.nvidia.com/XFree86/Linux-x86_64/450.57/README/dynamicpowermanagement.html
<mdnavare> and that says that they have some hacky way may be to power down the dGPU when not in use
<mdnavare> airlied: On a different note, how can we validate after setting DRI_PRIME=1 that the dGPU is being used for render
<mdnavare> ?
<mdnavare> other than adding debug prints in Mesa loader where it picks dGPU based on the DRI PRIME setting?
itoral has joined #dri-devel
<soreau> airlied: ha, that's a good one
<soreau> good song, too
sagar has left #dri-devel [#dri-devel]
sagar_ has joined #dri-devel
mlankhorst has joined #dri-devel
Anorelsan has joined #dri-devel
danvet has joined #dri-devel
NiksDev has joined #dri-devel
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
frieder has joined #dri-devel
Daanct12 has joined #dri-devel
pnowack has joined #dri-devel
pnowack has quit [Remote host closed the connection]
pnowack has joined #dri-devel
dos11 has joined #dri-devel
Daanct12 has quit [Read error: Connection reset by peer]
dos1 has quit [Read error: Connection reset by peer]
Daanct12 has joined #dri-devel
Duke`` has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
pnowack has quit [Quit: pnowack]
pochu has joined #dri-devel
camus1 has joined #dri-devel
camus has quit [Ping timeout: 480 seconds]
<danvet> melissawen, I've heard you care about v3d, so can you pls give this series some review/testing? https://lore.kernel.org/dri-devel/20210702213815.2249499-1-daniel.vetter@ffwll.ch/
pnowack has joined #dri-devel
pnowack has quit []
pnowack has joined #dri-devel
pnowack has quit [Remote host closed the connection]
pnowack has joined #dri-devel
mixfix41 has joined #dri-devel
mixfix41_ has quit [Ping timeout: 480 seconds]
<melissawen> danvet, sure
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #dri-devel
Daanct12 has quit [Remote host closed the connection]
lynxeye has joined #dri-devel
Daanct12 has joined #dri-devel
pcercuei has joined #dri-devel
Daanct12 has quit [Read error: Connection reset by peer]
pnowack has quit [Quit: pnowack]
pnowack has joined #dri-devel
rasterman has joined #dri-devel
pnowack has quit []
pnowack has joined #dri-devel
Ahuj has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
Duke`` has joined #dri-devel
camus has joined #dri-devel
camus1 has quit [Ping timeout: 480 seconds]
macromorgan has quit [Remote host closed the connection]
Duke`` has quit [Ping timeout: 480 seconds]
Duke`` has joined #dri-devel
pnowack has quit [Quit: pnowack]
<airlied> mdnavare: which gpu is for rendering is a userspace decision so no sysfs file
<airlied> mdnavare: you could use lsof or look in proc/pid/fd to see what gpu is opened maybe
rasterman has quit [Quit: Gettin' stinky!]
flacks has quit [Quit: Quitter]
flacks has joined #dri-devel
egbert has joined #dri-devel
pnowack has joined #dri-devel
rasterman has joined #dri-devel
camus1 has joined #dri-devel
camus has quit [Ping timeout: 480 seconds]
vivijim has joined #dri-devel
khfeng has quit [Quit: Konversation terminated!]
Bennett has joined #dri-devel
dv_ has quit [Ping timeout: 480 seconds]
Anorelsan has quit [Quit: Leaving]
pjakobsson has quit [Remote host closed the connection]
itoral has quit [Remote host closed the connection]
<agd5f> mdnavare, I suspect they are mainly focused on HG laptops. Those usually have a solution for the hotplug issue using the EC on the platform and the console is usually bound to the iGPU
pochu has quit [Ping timeout: 480 seconds]
sdutt has joined #dri-devel
<jekstrand> anholt: Are we disabling dEQP image logging for CI runs? We should. It substantially cuts disk I/O and probably helps CPU a bit too.
sneil_ has joined #dri-devel
sneil has quit [Ping timeout: 480 seconds]
sneil has joined #dri-devel
sneil_ has quit [Remote host closed the connection]
camus1 has quit []
aravind has quit []
pochu has joined #dri-devel
sneil_ has joined #dri-devel
sneil__ has joined #dri-devel
sneil_ has quit [Read error: Connection reset by peer]
sneil has quit [Ping timeout: 480 seconds]
pnowack has quit [Quit: pnowack]
Tooniis[m] has joined #dri-devel
pnowack has joined #dri-devel
dv_ has joined #dri-devel
mlankhorst has quit [Ping timeout: 480 seconds]
NiksDev2 has joined #dri-devel
NiksDev has quit [Remote host closed the connection]
Ahuj has quit [Ping timeout: 480 seconds]
heat has joined #dri-devel
tzimmermann has quit [Quit: Leaving]
mlankhorst has joined #dri-devel
GloriousEggroll has joined #dri-devel
macromorgan has joined #dri-devel
lynxeye has quit [Quit: Leaving.]
gouchi has joined #dri-devel
apteryx_ has joined #dri-devel
gouchi has quit []
bcarvalho has quit [Remote host closed the connection]
bcarvalho has joined #dri-devel
ngcortes has joined #dri-devel
mbrost has joined #dri-devel
pochu has quit [Read error: No route to host]
rsalvaterra has quit [Ping timeout: 480 seconds]
pochu has joined #dri-devel
rsalvaterra has joined #dri-devel
lanodan has quit [Quit: WeeChat 3.1]
lanodan has joined #dri-devel
Eighth_Doctor has joined #dri-devel
<Eighth_Doctor> hey all
<Eighth_Doctor> I'm not sure if I found the right place
<Eighth_Doctor> but I saw that https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9902 was merged
<Eighth_Doctor> does anyone know when a mesa release with that will be made available?
<pendingchaos> the first 21.2 release candidate should happen on the 14th
tobiasjakobi has joined #dri-devel
tobiasjakobi has quit [Remote host closed the connection]
dcbaker has joined #dri-devel
pochu has quit [Ping timeout: 480 seconds]
<kisak> pendingchaos: maybe? there isn't a volunteer to be a release maintainer for 21.2 yet
<pendingchaos> that's true
<dcbaker> I'm going to guess that my messages haven't been going through?
<dcbaker> In any case I'm planning to do the 21.2 release
<neobrain> dcbaker: yeah, the messages before "I'm going to guess" only went through on the matrix side, not on IRC. (Really wish the bridge weren't just silent about that)
<dcbaker> Apparently it failed to re-authenticate me with OFTC automatically and I had to go manually do it
<dcbaker> it being matrix
<Eighth_Doctor> well, I'm looking forward to the new Mesa RC coming out next week :)
<jenatali> dcbaker: Yeah I've noticed that too, whatever was set up for Freenode doesn't seem to work here. Fortunately it doesn't disconnect daily like Freenode did...
cmarcelo has joined #dri-devel
<cmarcelo> jenatali: was talking to jekstrand last week about nir_var_mem_image. the current idea is: use nir_var_mem_image for "storage images", keep using nir_var_uniform for textures. in OpenCL, we would keep them initially as nir_var_mem_images (since we don't have information on the type itself to say is not a var_mem_image) and then later lower to nir_var_uniforms the readonly ones... how does that look?
<jenatali> Caio Oliveira: About the same :)
<jenatali> If this is what people think is right then I'm fine with it, we'll just need to adapt - I can probably make the changes to our code to unblock you, but probably not until late next week, I'll be on vacation for a bit
hch12907 has quit [Read error: Connection reset by peer]
<karolherbst> anholt: so uhm.. this quarter I have time to work on the format stuff to fix it for big endian
<anholt> I think next step would be fixing the lp tests and llvmpipe failures, then one could hopefully fix the strange swizzle handling.
<karolherbst> yeah...
<karolherbst> I'd just remove this swizzle handling :D
<karolherbst> at least this "big endian" stuff in the table
<karolherbst> or is that already gone?
<anholt> I already did that, but there's still weird stuff in the .py (with big comments), and the different swizzles in the generated descs.
<karolherbst> ahh, cool
<karolherbst> okay.. I am nearly done with the nouveau mt stuff, so I think beginning next week or so I will start to update my knowledge about all of this
bcarvalho has quit [Remote host closed the connection]
bcarvalho has joined #dri-devel
moa has joined #dri-devel
mbrost has quit [Remote host closed the connection]
bluebugs is now known as Guest125
moa is now known as bluebugs
Guest125 has quit [Ping timeout: 480 seconds]
sagar_ has quit [Remote host closed the connection]
sagar_ has joined #dri-devel
sagar_ has quit [Remote host closed the connection]
sagar_ has joined #dri-devel
frieder has quit [Remote host closed the connection]
mbrost has joined #dri-devel
mbrost has quit [Read error: Connection reset by peer]
rasterman has quit [Quit: Gettin' stinky!]
mbrost has joined #dri-devel
xexaxo has joined #dri-devel
xexaxo has quit [Remote host closed the connection]
xexaxo has joined #dri-devel
<jekstrand> karolherbst: How are CL images coming? :P
<karolherbst> ugh...
xexaxo has quit [Remote host closed the connection]
xexaxo has joined #dri-devel
<karolherbst> I was thinking I like big huge topics, so I am working on those for now :D
<karolherbst> yeah.. dunno.. I think the patches are ready?
<jekstrand> And what about rustCL?
<jekstrand> :P
* jekstrand is just poking 'cause it's fun
<karolherbst> ehh.. let me check my next day of learning :D
<karolherbst> 27th of august :D
<karolherbst> jekstrand: yeah.. although one could potentially extract my core changes and do the general rust support stuff? :D
<karolherbst> dunno
<karolherbst> but my next step was to wire up the compiler and everything
<karolherbst> which includes nir
<karolherbst> jekstrand: if you want you can also work on it :D
GloriousEggroll has quit [Ping timeout: 480 seconds]
<jekstrand> karolherbst: Uh...
* jekstrand runs
<karolherbst> :D
<karolherbst> jekstrand: wasn't it you who tried to reimplement nir or some backend compiler in rust or something? :P
<jekstrand> karolherbst: I did play with compilers in rust a bit
<airlied> dcbaker: ping cant see you on other relvant chans, how far did you get with crocus va? and what were you testing with
<karolherbst> yeah.. although I will just generate bindings and be done with it
Thymo has quit [Quit: ZNC - http://znc.in]
GloriousEggroll has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
<dcbaker> airlied: I have a branch on my fdo gitlab
<dcbaker> I'm far enough that I need to figure out where the heck the galluim API is stuffing the reference frames
<dcbaker> for h264
<dcbaker> which sorta became a frustrating treck between the undocumented/uncommented gallium video layer, gallium va frontend, and libva
<dcbaker> I think I have most of the h264 decode state packets filled out correctly, minus setting up the BO's
<dcbaker> only on HSW, since that's what I have
<airlied> dcbaker: how are you testing it? i always fail at making a video ppayer work
<dcbaker> There's an assert somewhere early on that I hadn't fully tracked down, but I think is due to still lying a little bit about the video resources
<dcbaker> I'm still pretty fuzzy about how all of the gallium BOs work, and what the mapping between gallium concepts (which have 3d names) map to the video names
<dcbaker> like, what's a pipe_texture to the video API?
iive has joined #dri-devel
<airlied> dcbaker: I guess pipe_texture is where the output ends up
Thymo has joined #dri-devel
sdutt has quit []
sdutt has joined #dri-devel
Lucretia has quit []
Lucretia has joined #dri-devel
mlankhorst has quit [Ping timeout: 480 seconds]
<mdnavare> agd5f: Okay will need to check then how this is handled for Intel HG laptops
danvet has quit [Ping timeout: 480 seconds]
Bennett has quit [Remote host closed the connection]
bcarvalho has quit [Remote host closed the connection]
bcarvalho has joined #dri-devel
<mareko> dcbaker: you can use vl_video_buffer_create_as_resource to create pipe_video_buffer as a wrapper around pipe_resources
<mareko> dcbaker: then resource_create should be able to create planar textures where the current plane is pipe_resource and the next plane is in pipe_resource::next
<dcbaker> mareko: oooh, yeah, that makes sense
pnowack has quit [Quit: pnowack]
<pinchartl> stupid question: DRM defines a R8 and a R16 format, but no R10 or R12. is there no device that support 10- or 12-bit greyscale data ?
<pinchartl> (that is, with each pixel stored in 2 bytes, but shifted to the LSB, with 6 or 4 0's of padding on the MSB)
iive has quit []
pcercuei has quit [Quit: dodo]
heat has quit [Remote host closed the connection]
<jekstrand> anholt: As far as I'm concerned, if you think we've got consensus on ::name, marge it.
sagar_ has quit [Remote host closed the connection]
<jekstrand> anholt: So far we've got my patch, two RBs, and "Sure, not that useful" from krh, hakzsam, and Kayden.
sagar_ has joined #dri-devel
<jekstrand> Or I can marge it