ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
pcercuei_ has joined #dri-devel
pcercuei has quit [Ping timeout: 480 seconds]
pushqrdx has joined #dri-devel
pcercuei_ has quit []
pushqrdx has quit [Remote host closed the connection]
pushqrdx has joined #dri-devel
pushqrdx has quit [Remote host closed the connection]
pushqrdx has joined #dri-devel
co1umbarius has joined #dri-devel
columbarius has quit [Ping timeout: 480 seconds]
alyssa has quit [Quit: leaving]
pushqrdx has quit [Remote host closed the connection]
pushqrdx has joined #dri-devel
Net147 has quit [Quit: Quit]
Net147 has joined #dri-devel
Net147 has quit [Quit: Quit]
Net147 has joined #dri-devel
Net147 has quit [Remote host closed the connection]
Net147 has joined #dri-devel
i-garrison has joined #dri-devel
Net147 has quit [Quit: Quit]
Net147 has joined #dri-devel
Net147 has quit []
Net147 has joined #dri-devel
tarceri_ has quit [Remote host closed the connection]
JohnnyonFlame has quit [Read error: Connection reset by peer]
stuartsummers has quit [Ping timeout: 480 seconds]
unerlige has quit [Ping timeout: 480 seconds]
stuartsummers has joined #dri-devel
unerlige has joined #dri-devel
Duke`` has joined #dri-devel
lemonzest has joined #dri-devel
mlankhorst has joined #dri-devel
<HdkR>
Hm. Might need to send a patch for drirc. Looks like it might not like being loaded from sshfs
<HdkR>
I'm probably one of the only people loading drirc off of sshfs :D
gouchi has joined #dri-devel
YuGiOhJCJ has joined #dri-devel
vyivel has quit [Remote host closed the connection]
vyivel has joined #dri-devel
rasterman has joined #dri-devel
xexaxo_ has quit [Ping timeout: 480 seconds]
ced117 has quit [Ping timeout: 480 seconds]
ced117 has joined #dri-devel
<neobrain>
Hey, are DRI2/DRI3 expected to be available in a VM without graphics acceleration? I'm trying to run content with llvmpipe/lavapipe for reasons, but it looks like Vulkan fails to initialize due to DRI3 not being available
<sylware>
I guess this is already reported and handled
<sylware>
I guess I'll use next week git radv then
<sylware>
:)
<bnieuwenhuizen>
sylware: which GPU?
<bnieuwenhuizen>
also 5600 (xt)?
<sylware>
nav10 = 5600 XT
iive has joined #dri-devel
<sylware>
btw, officially, dota2 is dropping GL in the coming weeks.
<bnieuwenhuizen>
FWIW the problem shouldn't happen on 5700 xt which is also navi10
<sylware>
My bad, I thought 5700XT was not navi10
pushqrdx has quit [Remote host closed the connection]
pushqrdx has joined #dri-devel
<sylware>
side question: can we start looking to drop llvm for radv and use only aco?
<bnieuwenhuizen>
sylware: too late
<bnieuwenhuizen>
you can already build radv without llvm
<sylware>
DOH!
<sylware>
:)
<sylware>
I'll move to radv without llvm very soon then.
<bnieuwenhuizen>
note that radeonsi still needs it
<sylware>
because x11 is using GL, or there is some deeper interaction?
kts has joined #dri-devel
<bnieuwenhuizen>
no, just "if you want radeonsi you need GL", whether you want radeonsi is up to you
<FLHerne>
but although 'X11' doesn't require GL, lots of X11 applications and most DEs do
<FLHerne>
or at least disable their pretty window effects when it isn't available
<FLHerne>
and of course llvmpipe as a software fallback makes no sense with that complaint and softpipe is extremely slow
rasterman has joined #dri-devel
<sylware>
hu? There is a way to go native x11 without GL(accelerated or not) and use accelerated vulkan?
<FLHerne>
It's not about X11 [or Wayland] itself, but applications and desktops
vivijim has joined #dri-devel
<FLHerne>
Your favourite tiling WM will likely work fine without GL, but Firefox won't be able to render the nice 3D version of Google Maps, Krita will be very slow, the KDE desktop will look ugly with all the transparency effects disabled and so on
<FLHerne>
All of those *could* use Vulkan in principle, but they don't :p
<FLHerne>
You could use Zink + Penny at some point when the latter gets merged
<FLHerne>
(OpenGL over Vulkan)
garrison has joined #dri-devel
i-garrison has quit [Read error: Connection reset by peer]
<sylware>
I know, but could I have a xserver(xfree86 modesetting driver), this server not GL accelerated (hw or sw), and still have hw accelerated vulkan?
<Venemo>
sylware: btw a bisect is still appreciated if you can do it
<Venemo>
We don't have a 5600XT
<bnieuwenhuizen>
Venemo: the bug was bisected no?
<Venemo>
Then I'm out of the loop, sorrry
i-garrison has joined #dri-devel
garrison has quit [Read error: Connection reset by peer]
<FLHerne>
sylware: Yes
<FLHerne>
Just making sure it's clear that lacking OpenGL will affect more things than games
i-garrison has quit []
i-garrison has joined #dri-devel
<sylware>
Then, I guess, the xserver will be very slow blending the pixels buffers from the windows into the framebuffer.
pushqrdx has quit [Remote host closed the connection]
<sylware>
(since everything would be done on the CPU)
vivijim has quit [Ping timeout: 480 seconds]
gawin has joined #dri-devel
camus has quit []
pushqrdx has joined #dri-devel
Ahuj has joined #dri-devel
Anorelsan has joined #dri-devel
mlankhorst has quit [Ping timeout: 480 seconds]
kts has quit [Quit: Konversation terminated!]
mlankhorst has joined #dri-devel
soreau has quit [Read error: Connection reset by peer]
soreau has joined #dri-devel
flacks has quit [Ping timeout: 480 seconds]
sylware has quit [Remote host closed the connection]
dviola has quit [Remote host closed the connection]
dviola has joined #dri-devel
JohnnyonFlame has joined #dri-devel
flacks has joined #dri-devel
ppascher has quit [Ping timeout: 480 seconds]
gawin has quit [Ping timeout: 480 seconds]
Duke`` has quit [Ping timeout: 480 seconds]
gawin has joined #dri-devel
<kisak>
regarding llvm in my Ubuntu PPA, build farm restrictions prevent me from transitioning to llvm 13 for Focal and Hirsute, but I could transition to it with Bionic and Impish now.
<kisak>
a delay regularly happens with new major llvm version bumps.
Duke`` has joined #dri-devel
<kisak>
is there a mesa dev preference for me to I keep all the various mesa builds linking against the same llvm version or should I update sooner on the builds that can move forward?
ppascher has joined #dri-devel
<kisak>
mareko / pepp: maybe one of you have an opinion since radeonsi would be most affected
Ahuj has quit [Ping timeout: 480 seconds]
Anorelsan has quit [Quit: Leaving]
Ahuj has joined #dri-devel
lemonzest has quit [Quit: \]
gouchi has quit [Remote host closed the connection]
pushqrdx has quit [Remote host closed the connection]
pushqrdx has joined #dri-devel
gouchi has joined #dri-devel
ppascher has quit [Ping timeout: 480 seconds]
ppascher has joined #dri-devel
CME has quit [Ping timeout: 480 seconds]
<robclark>
karolherbst: out of curiousity, I perf-recorded some cl cts (test_basic hiloeo).. and it is nearly all in llvm/clang.. 87.5% in libclang-cpp, 7% in libLLVM... 0.4% in mesa.. I suppose kernels pre-compiled to spirv would help a lot. Not sure what the odds of duplicate shaders in a cts run are, but possibly clover disk-cache support would be a useful thing
<airlied>
there are some patches to reduce that overhead as well, I don't think you get too many duplicate shaders though
<airlied>
quite a lot of the clang overhead is processing opencl-c.h
<robclark>
I guess pre-compiled headers could be useful
<karolherbst>
robclark: ohh.. we should add a shader cache to clover I guess
<karolherbst>
but yeah..
<karolherbst>
maybe a pch for all the internal stuff would help
mlankhorst has quit [Ping timeout: 480 seconds]
CME has joined #dri-devel
<robclark>
looks like in hiloeo itself, there aren't any duplicate shaders.. but I guess a cache would help normal workloads quite a bit.. I guess I need something newer than f34 for airlied's patch to help
<robclark>
anyways, not a *big* deal yet, I guess I don't need too many tests to cover vec16, and kernel inputs and things like that.. load/store/atomic globals should get some coverage already in vk tests, I think
<airlied>
robclark: it might work on llvm13
<airlied>
I was just testing with latest as I was trying to test some fixes for the bugs
<robclark>
looks like llvm12 is what is in f34..
<robclark>
I guess I could move to f35.. just been lazy about that
<airlied>
yeah not so sure about that, only started investigating that feature
Duke`` has quit [Ping timeout: 480 seconds]
<karolherbst>
but I think we should pch the internal opencl header
<karolherbst>
I would be surprised if the compiler actually does much more than lexing the header
<karolherbst>
ehh parsing.. well both
<airlied>
the upstream answer is just to not use the header at all, and generate it internally in clang
<karolherbst>
huh?
<robclark>
is the clc->spv dependent at all on mesa version or driver caps? If not, we could do a separate cache for that which is populated at container build time, perhaps?
<airlied>
-fdeclare-opencl-builtins just generates the builtins from tablegen
<karolherbst>
okay
<karolherbst>
so your patch should speed up everything already
<airlied>
yup, once the last few llvm fixes land, I'll get to pushing it
* airlied
is trying to get cl3.0 + images compliant this week
<karolherbst>
cool
<karolherbst>
any issues left in my MR?
gouchi has quit [Remote host closed the connection]
* airlied
has one fix for kernel args parsing on top (outside of my hacks around images_used)
<airlied>
and CL3.0 needs read write images
<karolherbst>
right..
<karolherbst>
shouldn't be too complicated
<airlied>
also considering changing the CAP_IMAGES_SUPPORTED to return 2 values, one for read, one for write
<airlied>
rather than just reporting PIPE_*
<karolherbst>
welllll
<karolherbst>
ohh yeah
<karolherbst>
I guess it does make sense
<karolherbst>
read images are kind of textures, but CL woudn't allow that, right?
<karolherbst>
you can't just support read only images
<karolherbst>
mhh..
<karolherbst>
yeah, you have to support to write to images
<karolherbst>
I guess a _custom_ device won't have to
jewins has joined #dri-devel
<karolherbst>
airlied: do you have a use case for changing it to two values btw?
<airlied>
karolherbst: just they should be driver controlled numbers
<airlied>
not PIPE_*
<airlied>
since at least llvmpipe doesn't use those defines for its internal limits
<karolherbst>
ahh
gawin has quit [Ping timeout: 480 seconds]
Ahuj has quit [Ping timeout: 480 seconds]
jljusten has quit [Quit: WeeChat 3.3]
jljusten has joined #dri-devel
rauji_ has quit [Remote host closed the connection]
Company has quit [Quit: Leaving]
Lyude has quit [Quit: WeeChat 3.2]
Lyude has joined #dri-devel
qyliss has quit [Quit: bye]
pcercuei has quit [Quit: dodo]
qyliss has joined #dri-devel
qyliss has quit [Remote host closed the connection]
qyliss has joined #dri-devel
dreda has quit []
dreda has joined #dri-devel
iive has quit []
nsneck has quit [Remote host closed the connection]
alanc has quit [Remote host closed the connection]
rasterman has quit [Quit: Gettin' stinky!]
alanc has joined #dri-devel
<mareko>
kisak: updating sooner is better
jewins has quit [Ping timeout: 480 seconds]
kenjigashu has joined #dri-devel
kenjigashu has quit [Remote host closed the connection]