<dcbaker>
daniels: for all the things that stink about STL, its still good enough 99% of the time, and i didn't have to write it myself :)
smpl has quit [Quit: smpl]
isinyaaa has joined #dri-devel
isinyaaa has quit [Quit: WeeChat 3.2]
sdutt has quit [Ping timeout: 480 seconds]
blue__penquin has joined #dri-devel
camus has joined #dri-devel
Lightkey has quit [Ping timeout: 480 seconds]
Lightkey has joined #dri-devel
boistordu has joined #dri-devel
boistordu_ex has quit [Ping timeout: 480 seconds]
alatiera2 has joined #dri-devel
alatiera has quit [Ping timeout: 480 seconds]
khfeng has joined #dri-devel
<jekstrand>
dcbaker: Except basically every project that wants STL and wants perf ends up rewriting the STL.....
<airlied>
it's like everyone uses jemalloc :-P
<jekstrand>
I do wish we could use rust, though.
<HdkR>
airlied: I like my forked jemalloc thank you :)
<jekstrand>
I looked into what it'd take to tie Vulkan allocators into the rust standard lib once, though. Answer: Rewrite the rust standard lib. :-(
<jekstrand>
That or lots of TLS and weird hooking.
<jekstrand>
Which almost certainly wouldn't be reliable.
<airlied>
jekstrand: that's unfortunate, a vulkan driver would be a nice rust target
* zmike
idly mutters something about rewriting nir-to-spirv in rust
<jekstrand>
I mean, you could always just ignore that and use the rust standard lib anyway
<jekstrand>
I honestly don't know why we bother with the allocators
<jekstrand>
They do help catch leaks sometimes.
<jekstrand>
But 90% of the time, I think they exist to fix slow malloc() on windows
<airlied>
should propose making them removable :-P
<jekstrand>
As an extension!
<HdkR>
Doesn't the wine vulkan layer straight up ignore allocator arguments?
<jekstrand>
quite possibly
* orbea
waits for mesa to have a 7000 line cargo.lock file...
<jekstrand>
airlied: I think if I were to write a new Vulkan driver, I'd try to write it in Rust. I suspect I'd get about a month in before ditching rust and rewriting in C. :-/
<airlied>
dang no alyssa, was going to suggest tricking them into writing agx vk in rust :-P
<jekstrand>
A vk AGX driver in rust would be fun
* jekstrand
should start stubbing a vk rust driver as a w/e project just to see what it's like.
<jekstrand>
Don't need any hardware. Just build up a dispatch table, fill it with stuff, make some VkDevice etc. objects.
<zmike>
flakypipe
<dcbaker>
Ragepipe
<zmike>
^that's the one
<dcbaker>
Or maybe "damn it rustc your know what i want just do it"pipe
<zmike>
"BORROW CHECKER !@#$!@#$"
<zmike>
-pipe
<jekstrand>
Eh. I don't expect the borrow checker would be too much trouble.
<jekstrand>
There's gonna be so much unsafe, it's rediculous.
<jekstrand>
And we'd probably need something like box for wrapping vk objects
luzipher__ has joined #dri-devel
<zmike>
from what I remember the rust ffi is pretty okay
<zmike>
though I guess it's likely to have changed since I last used it
<jekstrand>
I don't know that you'd need that much ffi
<zmike>
shrug
<zmike>
I did the app embedding interface for servo, this seems like a similar thing
<jekstrand>
Lots of autogenerated c<->rust interface code, though.
<zmike>
yep
<jekstrand>
I'm not quite sure how you'd build the Vk entrypoint setup stuff to make the borrow checker useful.
<jekstrand>
Vulkan does have a reasonably well-defined concept of when an object is externally synchronized
* airlied
is more interested in some sort of rust/sycl hybrid for writing kernels :-P
<jekstrand>
Which could map reasonably well to &mut
<airlied>
great, though that's more for graphics kernels, would like one for SYCL type applications
luzipher_ has quit [Ping timeout: 480 seconds]
<jekstrand>
Rust is actually a pretty good match to shaders, IMO. All its push towards sticking everything on the stack matches really well.
<jekstrand>
airlied: Sure. But hopefully, one can build the one on the other.
<jekstrand>
I was a little surprised when they told me they were starting with 3D shaders and not compute. It does seem like compute would be easier.
sumits has joined #dri-devel
alatiera2 has quit [Ping timeout: 480 seconds]
<airlied>
jekstrand: yeah I just think having something like SYCL but not for C++ would be a good thing
camus1 has joined #dri-devel
camus has quit [Read error: Connection reset by peer]
aravind has joined #dri-devel
gpoo has quit [Ping timeout: 480 seconds]
Duke`` has joined #dri-devel
lemonzest has joined #dri-devel
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
orbea1 has joined #dri-devel
orbea has quit [Ping timeout: 480 seconds]
orbea1 has quit []
orbea has joined #dri-devel
mwalle has joined #dri-devel
itoral has joined #dri-devel
pnowack has joined #dri-devel
ZeZu has quit [Ping timeout: 480 seconds]
luzipher_ has joined #dri-devel
tzimmermann has joined #dri-devel
luzipher__ has quit [Ping timeout: 480 seconds]
bcarvalho has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
andrey-konovalov has joined #dri-devel
danvet has joined #dri-devel
tzimmermann has quit [Quit: Leaving]
tzimmermann has joined #dri-devel
Terman has quit [Remote host closed the connection]
rasterman has joined #dri-devel
rgallaispou has joined #dri-devel
daleh00k has joined #dri-devel
lynxeye has joined #dri-devel
dj-death has quit [Quit: leaving]
dj-death has joined #dri-devel
itoral_ has joined #dri-devel
itoral has quit [Read error: Connection reset by peer]
mlankhorst has joined #dri-devel
camus has joined #dri-devel
libv_ is now known as libv
thellstrom has joined #dri-devel
thellstrom1 has joined #dri-devel
camus1 has quit [Ping timeout: 480 seconds]
thellstrom has quit [Ping timeout: 480 seconds]
reinist12 has joined #dri-devel
reinist12 has left #dri-devel [#dri-devel]
reinist12 has joined #dri-devel
<reinist12>
Even unsure as of now what's going on! Rust-gpu is not the one you need, https://github.com/gfx-rs/gfx this one is , it even has vulkan-icd to anything, but there are so many of those that open source to every API to gallium is just a matter of couple days of work, and more than 10libraries could be implemented with sourcecode on one go, they are all very easy to be called to any language.
<reinist12>
they are all quite enough well structured, that it is immediately seen round about , just looking the source tree, how to generate translators, i have a big list of those cross api-translators, and by the look each one should work, but i favored webgpu line, and this uses sokol as intermediate step, it has no commercial restrictions zlib and public apache, hence anyone could get away with binaries easily.
<reinist12>
rest of them are some clause BSD or MIT's depends what licenses you need for mesa, and they can be plugged in for dx11 and dx12 most notably.
<reinist12>
so sokol does everything except dx12, which translationt to opengl or vulkan is generated from dawn headers
daleh00k has quit [Ping timeout: 480 seconds]
<reinist12>
but as for dx11, slimdx to opendk and whatnot, but all the api for directX is generated fro SharpDX or slimdx to c or c++, however you want also
jernej_ has quit [Read error: Connection reset by peer]
leaonidas has quit [Remote host closed the connection]
khfeng has quit [Remote host closed the connection]
khfeng has joined #dri-devel
reinist12 has quit []
mripard_ has joined #dri-devel
mripard has quit [Ping timeout: 480 seconds]
pnowack has quit [Quit: pnowack]
pnowack has joined #dri-devel
pnowack has quit []
pnowack has joined #dri-devel
<ishitatsuyuki>
can anybody check if sqtt works right now? i'm getting trace files that does not contains any actual information, but wonder if I just broke my environment
manu_ has left #dri-devel [#dri-devel]
evadot has joined #dri-devel
<ishitatsuyuki>
Radeon Developer Panel/AMDVLK works, but when I switch to RADV with the environment flags it emits a "small" trace file (~500KB) which RGP says "There are no events in the loaded profile so there is nothing to optimize".
<bnieuwenhuizen>
what GPU?
<bnieuwenhuizen>
have the oposite issue here where RADV works but RDP+amdgpu-pro does not :(
camus has quit [Remote host closed the connection]
camus has joined #dri-devel
aravind has quit []
<ishitatsuyuki>
5700 XT
<ishitatsuyuki>
you might need to set the Vulkan ICD if you haven't done it yet
itoral_ has quit []
camus has quit [Remote host closed the connection]
camus has joined #dri-devel
blue__penquin has quit []
pastly-antispam has quit [Quit: WeeChat 2.3]
gpoo has joined #dri-devel
khfeng has quit [Ping timeout: 480 seconds]
luzipher__ has joined #dri-devel
bcarvalho has quit [Read error: Connection reset by peer]
bcarvalho has joined #dri-devel
luzipher_ has quit [Ping timeout: 480 seconds]
<bl4ckb0ne>
emersion: do you have rights to assign PRs to marge?
Ryback_ has quit [Remote host closed the connection]
camus1 has joined #dri-devel
camus has quit [Read error: Connection reset by peer]
<jekstrand>
bl4ckb0ne: No, he doesn't yet
<jekstrand>
bl4ckb0ne: Need something merged?
<jekstrand>
Though he should. emersion has 40 mesa commits these days
<jekstrand>
emersion: Please file a bug and CC me and I'll get you some merge access
<danvet>
mripard_, I just pushed a fix to drm-misc-fixes but forgot to add cc: stable before pushing ...
<danvet>
otoh it's a fairly theoretical race I think
<danvet>
but if you feel like doing a rebase, could add it
<mripard>
do you want it today, or should we just wait until thurdsay?
<danvet>
one of the locking/lifetime fixes needs another round
<danvet>
so let's wait
mbrost has joined #dri-devel
andrey-konovalov has quit [Ping timeout: 480 seconds]
andrey-konovalov has joined #dri-devel
bcarvalho has joined #dri-devel
<daniels>
jadahl: new idea - does it work if you invert the order of the query_image_by_resource_handle() and query_image_by_resource_param() calls inside dri2_query_image()?
<daniels>
(src/gallium/frontends/dri/dri2.c)
<jadahl>
daniels: :o
<jadahl>
1/1 mutter:core+mutter/native/kms / native-kms OK 0.62s
<jadahl>
daniels: it did!
<jadahl>
that and the PARAM_MODIFIER handling in lp_texture together makes it work
<bnieuwenhuizen>
not really on the other hand my stuff using it is not ready
<dj-death>
jekstrand: thanks, that was definitely an improvement
<bnieuwenhuizen>
if your code is adjusted I'll just land it I guess
<dj-death>
bnieuwenhuizen: yeah, feel free, I'll update the anv MR
tobiasjakobi has joined #dri-devel
<jadahl>
daniels: thanks a lot for that, my life just got easier
<daniels>
jadahl: no problem, but I do wonder how on earth this broke in the first place - does it work if you revert back to b77f43f2539ddac6e334feefaa69a43720f1d2bf ?
sdutt has quit []
sdutt has joined #dri-devel
<jadahl>
daniels: will try, rebuilding mesa, but will fry some rice while the computer chrunches the numbers
<daniels>
nice :)
gouchi has joined #dri-devel
khfeng has quit [Ping timeout: 480 seconds]
<jekstrand>
anholt, mupuf: Does anyone have a quick example CTS test that causes the restrict message?
<imirkin>
wait, does that mean i should get cake 2x per year?
<imirkin>
and, plus, i'm owed back-cakes?
<jekstrand>
imirkin: Yup. Time to start stuffing your face. :P
<imirkin>
woohoo!
<jekstrand>
zmike: "You're usually so precise that I got confused." Really? You need to learn to trust me less. :P
<jadahl>
daniels: that commit also hits the same assert
<zmike>
jekstrand: look, when I'm at the gym and have to pick between putting that blood to work doing deep, inferential reads of review comments or getting more reps in, I'm always gonna make the same choice every time
<jekstrand>
zmike: Or you could do like I do and not bother going to the gym. :P
<jekstrand>
Or you could be a sensible person and leave GitLab alone while at the gym.
<zmike>
no, no, the rage fuels my gains
<zmike>
it's a positively reinforcing cycle
<ccr>
"good, goooood, I can feeeel your anger .."
<ccr>
"it gives you strength"
<imirkin>
i prefer to work out by eating cake.
<jekstrand>
^
<imirkin>
and as i've just realized, with all the back-cakes i'm owed for the half-year anniversaries, i'm gonna be doing a very heavy workout...
<bl4ckb0ne>
you can reward your workout by a cake
<imirkin>
good idea!
lemonzest has quit [Quit: Quitting]
<imirkin>
important to reward yourself after a good workout
<bl4ckb0ne>
otherwise what's the point of working out eh
<imirkin>
exactly
<imirkin>
(the good ol' eat-to-live vs live-to-eat debate)
<zmike>
I'm currently rewarding myself with a quick bisect
<bl4ckb0ne>
good reward
<imirkin>
zmike: i.e. if it goes long, you'll ignore the issue?
lynxeye has quit [Quit: Leaving.]
<imirkin>
hence the reward?
<zmike>
now you're getting it
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
<jekstrand>
Hey, look at all those NIR opcode docs!
<daniels>
jadahl: ah, I think it was broken by 2dcc9c7f54edbff075665ebe5d50f2499dc12163
<danylo>
zmike: does zink uses 5 descriptor sets? Turnip supports only 4...
<zmike>
yea I thought the min was higher for some reason
<zmike>
this should only affect you if you're using shader images for now though?
<danylo>
seems so
<danylo>
I crash in drawoverhead test =)
<danylo>
* I crashed in drawoverhead test =)
<zmike>
yeah there's image binds there
<zmike>
I can probably do some compacting to reduce the odds that you'll hit it and then for other cases roll the push set into the ubo set...
<zmike>
this is gonna be gross and I promised airlied we were done with descriptors for a while
<zmike>
I think you'll have to live with it like this for a bit, it's not going to be a quick fix in any case
<zmike>
maybe just comment out the image cases in drawoverhead for the time being?
<danylo>
Yes, I think it could wait, since the only users at the moment are some ppl who try to run dx9 games on android... I don't think they hit this path.
<danylo>
Yes, I should comment it out for testing
pcercuei has joined #dri-devel
<zmike>
I have to think a bit on how I can do this without doing another huge rewrite
<jekstrand>
Is that a HW limit in turnip or is it just going low because other drivers do?
<danylo>
The comment in the code says that it is an HW limit
<danylo>
Actually the limit is 5, but we need 1 set for dynamic descriptors and input attachments
<jekstrand>
zmike: You know how long Khronos takes now. I'm pretty sure fixing it will be faster. :P
<zmike>
fixing it or landing a fix?
<jekstrand>
/* no comment */
<zmike>
/* TODO: document this */
ngcortes has quit [Quit: Leaving]
ngcortes has joined #dri-devel
pcercuei has quit [Quit: bbl]
zackr has quit [Remote host closed the connection]
<kisak>
/* There used to be an interesting comment here */
<Sachiel>
/* Temporary hack to be fixed in a later commit. (1994-04-27) */
<zmike>
sensiblechuckle.gif
bcarvalho has quit [Ping timeout: 480 seconds]
<jenatali>
I always enjoy reading comments in ancient code
<jadahl>
daniels: sounds likely. will you create an MR with the changes required to make it work?
<daniels>
jadahl: I can but I think I'll rework it out of an abundance of caution
<jadahl>
alright. i'll keep the diff I have here now, at least it makes things work as much as making the tests pass (which doesn't check theh output, just logical state)
idr has joined #dri-devel
tzimmermann has quit [Quit: Leaving]
rasterman has quit [Ping timeout: 480 seconds]
rasterman has joined #dri-devel
<airlied>
danvet: had I not ff it' already, though I did yesterday
<airlied>
danvet: was going to backmerge rc7 to next and take the last few -next pulls
<cwabbott>
danylo: zmike: well, i think we already transitioned input attachments to use the non-bindless (GL) samplers, in theory we could do the same for dynamic/push descriptors i think
<cwabbott>
and reclaim that fifth set
<cwabbott>
but that would probably also be annoying
<zmike>
still probably something I should account for at some point even if you do get me a 5th set
mbrost has quit [Ping timeout: 480 seconds]
<bl4ckb0ne>
emersion: can the entry in docs/features.txt be in the radv commit?
<bnieuwenhuizen>
bl4ckb0ne: we typically do it in the commit that sets the extension in the extension list in the driver to true/supported/conditionally supported
<bl4ckb0ne>
ack
camus has joined #dri-devel
mbrost has joined #dri-devel
camus1 has quit [Ping timeout: 480 seconds]
mbrost has quit [Remote host closed the connection]
<bl4ckb0ne>
bnieuwenhuizen: move the entry to docs/relnote or duplicate the entry to docs/relnote?
<bnieuwenhuizen>
you can keep the features.txt entry too
<bnieuwenhuizen>
it is not the same format though so duplicating would be ill advised :)
<mupuf>
We get SpvDecorationInvariant and SpvDecorationRestrict
<mupuf>
there might be more, but these two are very common
<jekstrand>
mupuf: Yup. I was running a relase build. /o\
<mupuf>
jekstrand: do you think people always make their CTS runs using release builds?
<jekstrand>
They should run with release CTS and debug driver.
<mupuf>
right
* mupuf
has been confused as to how this could have been lasting for so long. We have been seeing this message since at least october. It felt like it became worse recently though
sravn has quit []
Peste_Bubonica has joined #dri-devel
* mupuf
is logging off... going camping in his basement to avoid the 30°C bedroom under the roof
rasterman has quit [Quit: Gettin' stinky!]
<jekstrand>
mupuf: Well, part of it is that they've got hand copied+pasted SPIR-V that has restrict in the wrong place. :-(
thellstrom1 has quit []
thellstrom has joined #dri-devel
sravn has joined #dri-devel
sravn has quit [Quit: WeeChat 2.9]
sravn has joined #dri-devel
rasterman has joined #dri-devel
sravn has quit []
sravn has joined #dri-devel
sravn has quit []
sravn has joined #dri-devel
sravn has quit []
<jekstrand>
mupuf: I took a look at the GLSLang code again now that I found a test case that fails.
<jekstrand>
mupuf: It's going to be a bit annoying to fix
<jekstrand>
So I'm not going to get a fix thrown together today like I was hoping.
* daniels
notes that what runs in CI is not at all magic but in fact something you can pull and run locally
<imirkin>
daniels: speaking of ... are there written-up docs (wiki pages) which explain how it all fits together?
<imirkin>
daniels: i had to pause my attempts at getting CI going locally a while back (due to an unfortunate misunderstanding of what was going wrong in grub), but i can pick it up now
<imirkin>
basically i switched to a usb keyboard, and it didn't seem to work in grub. i assumed the bios didn't support usb keyboards due to some setting that i had flipped. but no. it was just an old grub which relied on bios services. so i could access the bios all along but didn't realize it.
elongbug has quit [Remote host closed the connection]
sravn has joined #dri-devel
<daniels>
imirkin: that's quite good as far as CI problems go. just wait until you start having opinions about the longevity of physical relays, or the stability (& reliable power provision) of USB hubs and USB-TTL bridges, or even better the quality of the DHCP implementation on your fancy network-addressable power unit ...
<jekstrand>
mareko: Probably
mlankhorst has quit [Ping timeout: 480 seconds]
<bl4ckb0ne>
are the Xe DG1 available to public yet?
<bnieuwenhuizen>
bl4ckb0ne: for OEMs, yes
<HdkR>
In OEM systems yes
<jekstrand>
There are 3 laptops that have a hybrid TGL+DG!
<jekstrand>
DG1, rather
<jekstrand>
And I keep seeing people get their hands on DG1 PCI cards so I guess those are a thing
<imirkin>
daniels: i would like more longevity of the physical relays, and more stability of USB hubs / bridges, please!
<imirkin>
the more the merrier
<bl4ckb0ne>
cant find one on ebay yet
<imirkin>
is someone arguing for less? :)
<jekstrand>
bl4ckb0ne: You don't want oe
<jekstrand>
*one
<bnieuwenhuizen>
bl4ckb0ne: I think doing it that way is probably messy because you need motherboard support for it
<HdkR>
Also OEM business machine buyers typically take some time before parting out those machines :P
<bnieuwenhuizen>
jekstrand: if the mobo issues wasn't there it is actually quite decent as debugging vehicle without having to put together an entire computer?
<imirkin>
daniels: anyways, my problem was i couldn't boot a new kernel since i wouldn't be sure i could boot to another one from grub
<imirkin>
and i needed a new kernel for some of the podman/etc requirements
<bnieuwenhuizen>
then again until VRAM support in the kernel driver I supose it is meh
gouchi has quit [Remote host closed the connection]
<jekstrand>
bnieuwenhuizen: Probably. I'm more refering to the fact that your average user can't do anything with it right now.
<jekstrand>
I guess we're supposed to have some amount of i915 support landed soon.
<mareko>
it would be fine as a debug vehicle, though DG2 would be preferrable for debug + daily use
<jekstrand>
Still, not really worth it ATM.
<jekstrand>
Wait for a DG2
* bl4ckb0ne
waits
<imirkin>
jekstrand: always tempting to wait for N+1 :)
<jekstrand>
Seriously, the wait is worth it
<imirkin>
:)
<bl4ckb0ne>
:)
<imirkin>
don't worry, i'm not waiting in line for one
<airlied>
yeah the DG1 I have I think requrirs it's bios comes from the motherboard bios
<imirkin>
or any piece of hardware for that matter
<Venemo>
jekstrand: does DG stand for simply discrete graphics?
<jekstrand>
Venemo: No, it stands for something substantially more stupid.
<Venemo>
I'm listening
<jekstrand>
I'm not going to emberass my employer by admitting what. :-P
<daniels>
can it really be worse than New Intermediate Representation? or, like ... Windows New Technology
<Venemo>
jekstrand: I'm surprised it's not a lake or some other geographic place
<jekstrand>
mareko: Have you seen us try to use numbers?!?
<mareko>
I wish intel started using fish names too, to add extra confusion
<mareko>
jekstrand: sadly yes
<bnieuwenhuizen>
aren
<bnieuwenhuizen>
't two GFX enumas bad enough?
<Venemo>
mareko: bonus point if they use the same fish but a different color?
<mareko>
no, that's perfect obfuscation
<zmike>
what if intel did saltwater fish and amd did freshwater?
<DrNick>
Draw Good
<Sachiel>
that would set the bar too high
soreau has quit [Remote host closed the connection]
<DrNick>
Draw Gooder
soreau has joined #dri-devel
<Venemo>
Dinosaur Graphics? :P
ngcortes has joined #dri-devel
<mareko>
you need a cool new name, like RDNA = Radeon Destroys Nvidia Architecture
<zmike>
hahahahaha
<imirkin>
is that like NTSC = Never Twice the Same Color?
<HdkR>
DG = Damn Good
<HdkR>
It has to be
<HdkR>
:D
<daniels>
mareko: :D
<Venemo>
:)
rpigott has joined #dri-devel
YuGiOhJCJ has joined #dri-devel
camus1 has joined #dri-devel
camus has quit [Read error: Connection reset by peer]
idr has quit [Ping timeout: 480 seconds]
mmind00 has quit [Quit: mmind00]
andrey-konovalov has quit [Ping timeout: 480 seconds]
tobiasjakobi has quit [Remote host closed the connection]
rasterman has quit [Quit: Gettin' stinky!]
<anholt>
mareko: do you have pleasant scripts for comparing perf changes across all the drawoverhead tests?
tango_ has quit [Ping timeout: 480 seconds]
agx_ has quit [Ping timeout: 480 seconds]
mmind00 has joined #dri-devel
tango_ has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
iive has quit []
<mareko>
anholt: no, I used drawoverhead -test 11 -duration 4
<mareko>
or -duration 100 for sysprof
<anholt>
ok. I've got a freedreno branch that could use comparison across all the subtests because it plausibly changes them all
<mareko>
the sampler changes will be split into 3 MRs, 2 of them are public
<robclark>
hmm, is there some nir_validate vs constructing phi magic I'm missing.. getting nir_validate complaining validating phi src's but it kinda looks right to me:
<ParkerR>
Oh its still up over there I guess got claimed
<Sachiel>
lots of channels got registered in both places just in case
<ParkerR>
Well onto what I tried to ask there but its muted... heh
<ParkerR>
So Ive been trying linux git and linux next with the FreeSync HDMI patches from AMD (best I can tell they are already upstream but Im not 100% sure). Some of the patches mention DMCU HDMI FreeSync support but the cards I have are either too old (Polaris) or Raven where the dmcu firmware loading was patched out for the time being. When I try an HDMI display on either of those cards it never reports vrr_capable with the AMDGPU Xorg driver. Is
<ParkerR>
there some extra setup needed besides the Xorg conf VariableRefresh set to true? (Testing on linux-next 5.13-rc7)
camus has joined #dri-devel
rsripada_ has joined #dri-devel
aswar002_ has joined #dri-devel
mdnavare_ has joined #dri-devel
sdutt_ has joined #dri-devel
unerlige1 has joined #dri-devel
mbrost_ has joined #dri-devel
nchery has quit [Remote host closed the connection]
dolphin` has joined #dri-devel
nchery has joined #dri-devel
dolphin is now known as Guest142
dolphin` is now known as dolphin
shankaru1 has joined #dri-devel
sdutt has quit [Remote host closed the connection]
mbrost has quit [Remote host closed the connection]
camus1 has quit [Ping timeout: 480 seconds]
shankaru has quit [Ping timeout: 480 seconds]
unerlige has quit [Ping timeout: 480 seconds]
mdnavare has quit [Ping timeout: 480 seconds]
Guest142 has quit [Ping timeout: 480 seconds]
aswar002 has quit [Ping timeout: 480 seconds]
rsripada has quit [Ping timeout: 480 seconds]
<Lightkey>
jekstrand: Desktop Gaming?
agx_ has joined #dri-devel
mbrost_ has quit [Remote host closed the connection]