<airlied>
just got magically closed, maybe some fw fix went out but I've no idea, I never got back to testing it
DPA has quit [Ping timeout: 480 seconds]
ngcortes has quit [Read error: Connection reset by peer]
<airlied>
there were many reasons to give up the sdma quest :-P
<jenatali>
I don't think anybody here is affected by this, but I'm looking for etiquette advice: If I have a major change like https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12677, how long should I wait for people to weigh in before going ahead?
<airlied>
jenatali: if nobody raises any major objections, a week is usually about what I'd give it
<bnieuwenhuizen>
jenatali: I think the last strongly worded thread about this was "at least 24 hours" (on workdays I assume). Longer is fine of course if you think it is beneficial but don't think there has been something where we discussed it till consensus
<jenatali>
Cool, thanks, I was leaning towards a week but wasn't sure if (e.g.) vacations or stuff meant some changes should wait longer
<mareko>
jenatali: the number of wgl stakeholders is low and vmware replied already, so getting more feedback is unlikely
<jenatali>
mareko: Yep, good point, thanks
columbarius has joined #dri-devel
co1umbarius has quit [Ping timeout: 480 seconds]
DPA has joined #dri-devel
nchery has quit [Quit: Leaving]
pnowack has quit [Quit: pnowack]
<Plagman>
anyone aware of a vaapi hw encode test app somewhere?
<Plagman>
(for h264)
<kisak>
OBS studio?
<Plagman>
meant more minimal than that, does obs use ffmpeg? i could take a peek
<bnieuwenhuizen>
IIRC ffmpeg has vaapi encoding support
<imirkin_>
Plagman: ffmpeg should be able to from cmdline, no?
JohnnyonFlame has quit [Ping timeout: 480 seconds]
bfvwtdsl^ has quit [Remote host closed the connection]
mattrope has quit [Remote host closed the connection]
mattrope has joined #dri-devel
thellstrom1 has joined #dri-devel
thellstrom has quit [Read error: Connection reset by peer]
<Plagman>
yes
<Plagman>
commands in the link kisak posted are great for that
<Plagman>
i was more looking for test code than a test app, ffmpeg is fairly big
<Plagman>
but looking at it closer it seems like ffmpeg and using vaapi for h264 encoding might be somewhat intertwined anyhow, so looks like it's the place to start at any rate
<Plagman>
thanks
<imirkin_>
yeah, i don't think anything just spits out a plain h264 bitstream ... though maybe? dunno
<imirkin_>
(anything in the va-api that is)
camus1 has joined #dri-devel
camus has quit [Read error: Connection reset by peer]
idr has quit [Quit: Leaving]
Duke`` has joined #dri-devel
NiksDev has joined #dri-devel
vivijim has quit [Remote host closed the connection]
<tomeu>
anholt_: btw, I had some success with deqp-runner running the tests within crosvm by means of a wrapper script, but the renderer check is causing problems there, as it's run outside the VM
<tomeu>
I think I heard some mention of plans to move it into deqp-runner, is that right?
f11f12 has quit [Quit: Leaving]
sdutt has quit [Read error: Connection reset by peer]
lemonzest has joined #dri-devel
mattrope has quit [Remote host closed the connection]
Hi-Angel has quit [Read error: Connection reset by peer]
frieder has joined #dri-devel
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
pnowack has joined #dri-devel
gouchi has joined #dri-devel
gouchi has quit [Remote host closed the connection]
xexaxo_ has joined #dri-devel
jessica_24 has quit [Quit: Connection closed for inactivity]
sarnex_ has joined #dri-devel
sarnex has quit [Remote host closed the connection]
rasterman has joined #dri-devel
jkrzyszt has joined #dri-devel
ivyl has quit [Quit: end of flowers]
f11f12 has joined #dri-devel
danvet has joined #dri-devel
adavy has quit [Ping timeout: 480 seconds]
tursulin has joined #dri-devel
pcercuei has joined #dri-devel
rsalvaterra_ has quit []
ivyl has joined #dri-devel
rsalvaterra has joined #dri-devel
ivyl is now known as Guest6665
Guest6665 has quit []
ivyl_ has joined #dri-devel
ivyl_ has left #dri-devel [#dri-devel]
ivyl has joined #dri-devel
thellstrom has joined #dri-devel
thellstrom1 has quit [Ping timeout: 480 seconds]
shfil has joined #dri-devel
thellstrom has quit [Remote host closed the connection]
thellstrom has joined #dri-devel
Hi-Angel has joined #dri-devel
Danct12 has quit [Remote host closed the connection]
Danct12 has joined #dri-devel
xexaxo_ has quit [Read error: No route to host]
xexaxo_ has joined #dri-devel
Ahuj has joined #dri-devel
Ahuj has quit []
Ahuj has joined #dri-devel
xexaxo_ has quit [Read error: Connection reset by peer]
hanetzer3 has quit [Ping timeout: 480 seconds]
hanetzer has joined #dri-devel
rsalvaterra_ has joined #dri-devel
rsalvaterra has quit [Ping timeout: 480 seconds]
rsalvaterra_ has quit []
rsalvaterra has joined #dri-devel
<pq>
MrCooper, thanks for the explanation last night, I finally understand why async cursor updates could be useful. :-) IOW, a workaround until we can guarantee display (servers) won't get stalled by clients.
<MrCooper>
no worries
hanetzer1 has joined #dri-devel
hanetzer has quit [Ping timeout: 480 seconds]
hanetzer1 has quit [Ping timeout: 480 seconds]
robink has quit [Ping timeout: 480 seconds]
<daniels>
emersion, pq: thanks a lot both for the comments!
JohnnyonFlame has joined #dri-devel
<pq>
haai
adavy has joined #dri-devel
robink has joined #dri-devel
hanetzer has joined #dri-devel
jhli_ has quit [Ping timeout: 480 seconds]
hanetzer has quit [Remote host closed the connection]
hanetzer has joined #dri-devel
Company has joined #dri-devel
xexaxo_ has joined #dri-devel
vivijim has joined #dri-devel
hanetzer has quit [Read error: Connection reset by peer]
hanetzer has joined #dri-devel
adjtm is now known as Guest6680
adjtm has joined #dri-devel
hanetzer1 has joined #dri-devel
hanetzer is now known as Guest6681
hanetzer1 is now known as hanetzer
Guest6680 has quit [Ping timeout: 480 seconds]
Guest6681 has quit [Ping timeout: 480 seconds]
JohnnyonFlame has quit [Read error: Connection reset by peer]
soreau has quit [Remote host closed the connection]
gsn has joined #dri-devel
soreau has joined #dri-devel
i-garrison has quit []
i-garrison has joined #dri-devel
sdutt has joined #dri-devel
gsn has quit []
gsn has joined #dri-devel
gsn has quit []
sdutt has quit []
sdutt has joined #dri-devel
mattrope has joined #dri-devel
xexaxo_ has joined #dri-devel
<Company>
question: I've been looking at floating point texture support in GTK's pipeline - does Wayland or do GPUs care about channel order - RGBA vs BGRA vs whatever - and what's the opinion on premultiplied vs unpremultiplied colors?
<Company>
because Wayland's historically used premultiplied buffers, but it seems weird for floating point
<ajax>
there's only one common channel order in hardware for rgba16f and rgba32f, afaik
<Company>
ajax: and that's float[4] (or fp16[4]) in RGBA order?
<Company>
I haven't seen BE hardware in ages, so I would have no idea what's going on there
<MrCooper>
endianness affects the order of bytes in a multi-byte word, not the order of multi-byte words
<Company>
I'm just making sure
<Company>
people and hardware didn't agree on uint8_t vs uint32_t for colors, so maybe they don't agree on float vs vec4 either - myknowledge of special-case hardware is quite limited
<MrCooper>
8-bit per component formats are the special case which can be ambiguous vs endianness
<ajax>
Company: yes. index 0 is r, index 3 is a
<Company>
great
<Company>
so FP16_RGBA and FP32_RGBA are probably the only formats we want
<Company>
leaves the premultiplied question: I suppose premultiplied is not something we want for floats?
<ajax>
i'm not sure the question makes physical sense? what would an alpha > 1.0 mean
<zmike>
REALLY bright
<ajax>
i get what an overbright pixel might be but not an oversolid pixel
<Company>
it's mostly about if vec(0.5, 0.5, 0.5, 0.5) is semitransparent gray or semitransparent white
<ajax>
this pixel hit 17 of the 16 samples when rasterising
<zmike>
haha
<Company>
also, is 3 channels a thing worth supporting with floats?
<ajax>
no
<ajax>
at least: i don't think much hardware supports three-channel floats as a render or blit format
<MrCooper>
does Wayland currently assume pre-multiplied alpha, or did I dream that?
* ajax
consults vulkaninfo
<Company>
MrCooper: it does
<Company>
MrCooper: implicitly I think, nobody ever mentions it
<Company>
MrCooper: but people started Wayland by throwing Cairo surfaces at it, and those are premultiplied
<Company>
also, X is premultiplied, too
<Company>
but I don't think premultiplied makes much sense in a floating point world
<Company>
it's very neat with porter-duff compositing, but DEST_OUT is not really what people get excited about anymore
<MrCooper>
a penny for pq's thoughts on this
<ajax>
premul is an encoding trick for cpus, imao
<Company>
except that GTK is using it in its GL renderer, too - I'll need to find somebody to fix that
* Company
goes convincing mclasen
<pq>
I was just going to run home. :-P
<ishitatsuyuki>
what can be bad with premultiplied?
<ajax>
Company: random sampling of gpus i have handy. radeon rx480 and nvidia gtx1650 can kindasorta do rgb32f but not rgb16f. intel coffeelake has somewhat complete support for both.
<MrCooper>
pq: no rush then
<ajax>
Company: i'd say add support iff someone really needs the reduced storage or bandwidth and asks for it?
<Company>
ajax: a bunch of image formats seem to use it
<Company>
ajax: because it turns out photos don't really contain alpha channels
<ishitatsuyuki>
I guess rgba16f would make better sense as a fallback? not sure what's the topic though
<pq>
Company, FWIW, Wayland has no way to communicate "straight alpha" until we split that part from the CM&HDR protocol extension.
<ajax>
blah
<Company>
pq: I'm trying to look into the future to figure out the HDR protocol that everybody will settle down with
<pq>
Company, do not ever, ever, hardcode any assumptions from pixel format, like premult or sRGB. It only leads to a mess.
<Company>
pq: so that I can pick that as GTK's HDR format already
<ajax>
the only thing premul can do for you, as a transport format, is give you illegal pixel encodings and crush your color fidelity
<Company>
ajax: yeah, that was my assumption, too
<Company>
so I'm gonna predict that fp16 unpremultiplied is gonna win out
<pq>
premultiplied alpha does make everything harder, particularly in color management; except the one niche case of trivial blending on CPU. :-)
<pq>
however, Wayland is and remains premultiplied until we have a protocol extension to say otherwise.
<Company>
yeah, but wayland also remains U8 until that extension exists
<pq>
Company, HDR is not a format.
<pq>
Company, the only thing guaranteed is that Wayland compositors will accept premult. Straight alpha will be hugely preferred, but also not guaranteed.
<pq>
although, who wouldn't support straight alpha once the protocol is there...
<Company>
yeah
<Company>
we'll definitely have a way to end up with premult
<Company>
we need that for X anyway
<pq>
so... I think it's a safe bet to require straight alpha support from the window system with HDR.
<pq>
X won't have HDR, judging from its progress that I've seen.
<pq>
Company, nope, the CM&HDR protocol extension will not add float-point formats. Those are already in.
<Company>
oh
<Company>
and Wayland assumes those are premult?
<pq>
oh course, pixel format has nothing to do with premult
<pq>
*of
<Company>
just making sure
<Company>
so I'll push the GTK GL/Vulkan rendering code to go straight alpha because it's less pain and because that's what we'll want in a sane system anyway
<pq>
looks like even wl_shm defines 16F formats, and zwp_linux_dmabuf_v1 just says "use drm_fourcc.h"
<Company>
and do a postprocessing step that premults - and once the HDR stuff comes around, I'll remove that
<pq>
Company, well, all these are optional for compositors to support, so you probably don't want to remove it.
<Company>
yeah, remove if supported
<pq>
ah, in that sense, yeah
<jadahl>
pq: we sync the shm formats from fourcc, so everything there gets copied when anyone cares to run the sync scipt
<pq>
jadahl, I regret not just saying to use drm_fourcc.h there too.
<Company>
and I'll lobby jadahl or whoever writes the code for mutter/shell to also use fp16 straight
<pq>
premult is fun with color management, because the first thing a compositor must always do is undo that
<Company>
so that a Gnome desktop ends up either fp16 or U8 straight alpha from top to bottom
<jadahl>
pq: and the cairo one there is a bit of an odd ball
<pq>
jadahl, yeah - never ask what they should do on BE.
jessica_24 has joined #dri-devel
<jadahl>
whats the current support for using fp16 on the primary plane?
<MrCooper>
drm_info says it's supported with my AMD GPUs
<emersion>
XBGR16161616F seems most widely supported, but only for amdgpu/i915/nouveau
<emersion>
it's the number of devices that support the format, divided by the total number of devices in the DB
<jadahl>
ah, ok
<emersion>
well
<emersion>
that's not quite right
Duke`` has joined #dri-devel
<emersion>
if you pick the XBGR16161616F row and the amdgpu column it says 48%
<emersion>
48% of amdgpu devices support XBGR16161616F scanout
imirkin has joined #dri-devel
<emersion>
should probably add a tooltip which spells it out
<pq>
I suppose if your display connection happens to be carrying only 10 bpc, then 10 bpc might do on the FB - at least if you don't need the precision for KMS LUTs and CTM.
<ajax>
pq: i'm reasonably confident that the first X server to have HDR support would be Xwayland
nchery has joined #dri-devel
<pq>
ajax, isn't there a slight detail in the way? X11 and visuals? Or can you somehow skip that problem if the X server only forwards the... HDR... metadata? Where do you get the metadata?
<ajax>
we wrote a spec for how it should work, it probably still applies
<ajax>
i was more saying i wouldn't expect, like, xserver/hw/xfree86/ to grow support for it
<ajax>
under Xwayland it's trivial, i translate some enums and i add a new visual class and tripwire it away from all the legacy paths
<pq>
Right, if any X would get HDR support, it would be Xwayland, sure. But it's the "if" I'm thinking about. Who would use it?
<pq>
now I really need to go, till tomorrow \o.
Ahuj has quit [Ping timeout: 480 seconds]
<danvet>
pq, 10 bpc formats aren't gaining you much coverage
<danvet>
the arm-socs don't bother
<danvet>
or don't have it, dunno
<danvet>
maybe it's worth it for memory bw reasons as a fallaback, since it's half the size
mattrope has quit [Read error: Connection reset by peer]
* alyssa
working on her XDC slides
<alyssa>
I have like a week right? pshh
<Sachiel>
why are you doing it now then?
thellstrom has quit [Ping timeout: 480 seconds]
mattrope has joined #dri-devel
thellstrom has joined #dri-devel
<robclark>
danvet: I think the 10b format that will see more use is 10_10_10_2.. AFAIU that is what CrOS uses for HDR scanout (on x86 and arm)
gouchi has joined #dri-devel
<ajax>
does anyone do r11g11b10 as a display format?
<imirkin>
certainly not unorm. presumably you mean the float thing?
<danvet>
why is there a discussion going on on dri-devel about reinventing systemd-logind, but badly
<danvet>
robclark, do you have hw where 1010102 is supported, but not 16f?
<danvet>
for cros I expect the bw savings is justification enough, but maybe mutter doesn't care
<imirkin>
ajax: you could almost do it with 10_10_10 + LUT :)
<robclark>
danvet: I do as a matter of fact.. dpu supports 10_10_10_2 (in various swizzles)
<robclark>
but yeah, it's preferred since no bandwidth and memory footprint penalty
rgallaispou has left #dri-devel [#dri-devel]
<bnieuwenhuizen>
if you want more range there is also rgb9e5, some newer GPUs support that for rendering now, would be cool for display too (unless you need the alpha ...)
<emersion>
what is "e"?
<bnieuwenhuizen>
exponent
<bnieuwenhuizen>
basically 3 floats using a shared exponent
<imirkin>
shared-exponent float
<emersion>
ah interesting
<bnieuwenhuizen>
which gives you the ability to go really small or big but still keep 32-bits
<swick>
10bpc formats are mostly fine for scanout but not great for compositing
<swick>
so there is a tradeoff like always
<swick>
danvet: logind only passes around a DRM fd though?
<danvet>
swick, I thought it's doing the master handover too
jhli has joined #dri-devel
<danvet>
because you can't do that if you're not root
jkrzyszt has quit [Ping timeout: 480 seconds]
<danvet>
robclark, I didn't see that on emersion drminfo page
<danvet>
maybe upload a dump
<danvet>
or I'm blind
<robclark>
drminfo page?
<swick>
danvet: sure but nothing that makes handover any better
<robclark>
danvet, emersion, it might be a bug that sdm845 doesn't expose 10_10_10_2.. but IIRC some devices support it and others do not
ramaling has quit [Remote host closed the connection]
ramaling has joined #dri-devel
Ryback_ has quit [Remote host closed the connection]
jhli has quit [Remote host closed the connection]
rsripada has quit [Read error: Connection reset by peer]
nchery has quit [Remote host closed the connection]
aswar002 has quit [Remote host closed the connection]
unerlige has quit [Remote host closed the connection]
Ryback_ has joined #dri-devel
shankaru has quit [Remote host closed the connection]
pzanoni has quit [Remote host closed the connection]
mdnavare has quit [Remote host closed the connection]
sdutt has quit [Remote host closed the connection]
vivijim has quit [Remote host closed the connection]
mattrope has quit [Remote host closed the connection]
mdnavare has joined #dri-devel
pzanoni has joined #dri-devel
nchery has joined #dri-devel
vivijim has joined #dri-devel
jhli has joined #dri-devel
tursulin has quit [Remote host closed the connection]
sdutt has joined #dri-devel
dolphin has quit [Remote host closed the connection]
dolphin has joined #dri-devel
mattrope has joined #dri-devel
thellstrom has quit [Remote host closed the connection]
unerlige has joined #dri-devel
shankaru has joined #dri-devel
ngcortes has joined #dri-devel
aswar002 has joined #dri-devel
tursulin has joined #dri-devel
rsripada has joined #dri-devel
<alyssa>
danvet: Apple uses a two plane RGB101010 / A8 format
<alyssa>
Really not looking forward to piping that into userspace.
<alyssa>
no packed rgb10_a2 that I can see on the display side
mlankhorst has quit [Ping timeout: 480 seconds]
<danvet>
alyssa, oh for overlays?
<danvet>
and yeah argb2101010 is probably jarring in the transitions
<danvet>
it's only good for xrgb2101010 I guess
<bnieuwenhuizen>
or when the transition is 0/1 really
<danvet>
alyssa, wrt piping, shouldn't be any worse than any other multi-plane format
<bnieuwenhuizen>
would be decent for e.g. being trnasparent where one has an underlay
idr has joined #dri-devel
<robclark>
I guess 2 bits of alpha is probably good enough to do anti-aliased edges (like round corners)?
<danvet>
yeah round corners with retina might be good enough
<danvet>
if you do actual drop shadows, it'd be terrible
<danvet>
like 4xmsaa is also just 2 bits
jhli has quit [Remote host closed the connection]
Ryback_ has quit [Remote host closed the connection]
nchery has quit [Remote host closed the connection]
unerlige has quit [Remote host closed the connection]
aswar002 has quit [Remote host closed the connection]
Ryback_ has joined #dri-devel
sdutt has quit [Remote host closed the connection]
mdnavare has quit [Remote host closed the connection]
mdnavare has joined #dri-devel
ngcortes has quit [Remote host closed the connection]
jhli has joined #dri-devel
unerlige has joined #dri-devel
sdutt has joined #dri-devel
tursulin has quit [Remote host closed the connection]
ngcortes has joined #dri-devel
dolphin has quit [Remote host closed the connection]
heat has joined #dri-devel
ramaling has quit [Remote host closed the connection]
rsripada has quit [Remote host closed the connection]
rsripada has joined #dri-devel
ramaling has joined #dri-devel
tursulin has joined #dri-devel
shankaru has quit [Remote host closed the connection]
aswar002 has joined #dri-devel
pzanoni has quit [Remote host closed the connection]
pzanoni has joined #dri-devel
mattrope has quit [Remote host closed the connection]
vivijim has quit [Remote host closed the connection]
Ryback_ has quit [Remote host closed the connection]
jhli has quit [Remote host closed the connection]
vivijim has joined #dri-devel
unerlige has quit [Remote host closed the connection]
rsripada has quit [Remote host closed the connection]
aswar002 has quit [Remote host closed the connection]
jhli has joined #dri-devel
mdnavare has quit [Remote host closed the connection]
sdutt has quit [Remote host closed the connection]
ngcortes has quit [Remote host closed the connection]
mdnavare has joined #dri-devel
mattrope has joined #dri-devel
unerlige has joined #dri-devel
shankaru has joined #dri-devel
sdutt has joined #dri-devel
ngcortes has joined #dri-devel
aswar002 has joined #dri-devel
dolphin has joined #dri-devel
rsripada has joined #dri-devel
Ryback_ has joined #dri-devel
<alyssa>
danvet: macOS uses rgb10+a8 as the default primary plane format for HDR
<danvet>
emersion, I'm going to chicken out of reviewing your empty lease patches
<danvet>
mostly because I'm just burried :-(
<emersion>
;_;
<danvet>
but with igt and everything is fine with me imo
<danvet>
pls just make sure it does all land, we've had igts in the past that got lost
<danvet>
holler at ivyl or Adrinael if you cant get igt patches pushed
mbrost has joined #dri-devel
<emersion>
any idea who could review the kernel bits?
<emersion>
i don't think keith is too involved anymore
mdnavare has quit [Remote host closed the connection]
sdutt has quit [Remote host closed the connection]
Ryback_ has quit [Remote host closed the connection]
mbrost has quit [Remote host closed the connection]
jhli has quit [Remote host closed the connection]
Ryback_ has joined #dri-devel
mdnavare has joined #dri-devel
ramaling has quit [Remote host closed the connection]
ramaling has joined #dri-devel
mbrost has joined #dri-devel
jhli has joined #dri-devel
tursulin has quit [Remote host closed the connection]
dolphin has quit [Remote host closed the connection]
shankaru has quit [Remote host closed the connection]
dolphin has joined #dri-devel
pzanoni has quit [Remote host closed the connection]
pzanoni has joined #dri-devel
ngcortes has quit [Remote host closed the connection]
rsripada has quit [Remote host closed the connection]
unerlige has quit [Remote host closed the connection]
aswar002 has quit [Remote host closed the connection]
tursulin has joined #dri-devel
ngcortes has joined #dri-devel
vivijim has quit [Remote host closed the connection]
mattrope has quit [Remote host closed the connection]
vivijim has joined #dri-devel
unerlige has joined #dri-devel
mattrope has joined #dri-devel
shankaru has joined #dri-devel
rsripada has joined #dri-devel
ngcortes has quit [Remote host closed the connection]
jhli has quit [Remote host closed the connection]
Ryback_ has quit [Remote host closed the connection]
mbrost has quit [Remote host closed the connection]
mdnavare has quit [Remote host closed the connection]
aswar002 has joined #dri-devel
mripard has quit [Ping timeout: 480 seconds]
mdnavare has joined #dri-devel
ramaling has quit [Remote host closed the connection]
mbrost has joined #dri-devel
jhli has joined #dri-devel
dolphin has quit [Remote host closed the connection]
tursulin has quit [Remote host closed the connection]
dolphin has joined #dri-devel
bbrezillon has quit [Ping timeout: 480 seconds]
shankaru has quit [Remote host closed the connection]
pzanoni has quit [Remote host closed the connection]
pzanoni has joined #dri-devel
aswar002 has quit [Remote host closed the connection]
rsripada has quit [Remote host closed the connection]
unerlige has quit [Remote host closed the connection]
mattrope has quit [Remote host closed the connection]
vivijim has quit [Remote host closed the connection]
unerlige has joined #dri-devel
ngcortes has joined #dri-devel
mattrope has joined #dri-devel
shankaru has joined #dri-devel
tursulin has joined #dri-devel
aswar002 has joined #dri-devel
nchery has joined #dri-devel
bbrezillon has joined #dri-devel
Ryback_ has joined #dri-devel
ramaling has joined #dri-devel
rsripada has joined #dri-devel
vivijim has joined #dri-devel
mripard has joined #dri-devel
soreau has quit [Ping timeout: 480 seconds]
Peste_Bubonica has joined #dri-devel
tanty has joined #dri-devel
soreau has joined #dri-devel
Peste_Bubonica has quit [Quit: Leaving]
tanty has quit [Quit: Ciao!]
Company has quit [Ping timeout: 480 seconds]
Company has joined #dri-devel
Guest6630 has left #dri-devel [#dri-devel]
pepp has joined #dri-devel
JohnnyonFlame has joined #dri-devel
gouchi has quit [Remote host closed the connection]
gouchi has joined #dri-devel
gouchi has quit [Remote host closed the connection]
gouchi has joined #dri-devel
gouchi has quit [Remote host closed the connection]
gouchi has joined #dri-devel
lemonzest has quit [Quit: WeeChat 3.2]
JohnnyonFlame has quit [Remote host closed the connection]
frieder has quit [Remote host closed the connection]
Duke`` has quit [Ping timeout: 480 seconds]
<Company>
ajax, pq: so owen mentioned that it turns out unpremultiplied breaks bilinear filtering, so using unpremultiplied suddenly becomes a lot less appealing
JohnnyonFlame has joined #dri-devel
rasterman has quit [Quit: Gettin' stinky!]
oneforall2 has quit [Quit: Leaving]
vivijim has quit [Ping timeout: 480 seconds]
lemonzest has joined #dri-devel
gouchi has quit [Remote host closed the connection]
mbrost has quit [Read error: Connection reset by peer]
oneforall2 has joined #dri-devel
<mareko>
premultiplied alpha?
ppascher has quit [Ping timeout: 480 seconds]
<mareko>
it's certainly required for filtering and compositing because blending with premultiplied alpha is mathematically an associative operation
<mareko>
is 16f scanout going to be a thing anytime soon?
<imirkin>
it's been a thing since DX10-class nvidia hw...
heat has quit [Ping timeout: 480 seconds]
ngcortes has quit [Remote host closed the connection]
Hi-Angel has quit [Ping timeout: 480 seconds]
<airlied>
imirkin: I think more in the context of anyone using it
<bnieuwenhuizen>
kisak: all the more reason to encourage people to use 21.2 :)
<kisak>
I'll throw it into my stable mesa ppa for the 3 people who use it
danvet has quit [Ping timeout: 480 seconds]
bluebugs has quit []
iive has quit []
cedric has joined #dri-devel
cedric is now known as bluebugs
lemonzest has quit [Quit: WeeChat 3.2]
sdutt has joined #dri-devel
rasterman has joined #dri-devel
rcf has quit [Remote host closed the connection]
flto_ has joined #dri-devel
rcf has joined #dri-devel
flto has quit [Ping timeout: 480 seconds]
pcercuei has quit [Quit: dodo]
Surkow|laptop has quit [Remote host closed the connection]
Surkow|laptop has joined #dri-devel
flto_ has quit []
flto has joined #dri-devel
nchery has quit [Remote host closed the connection]
bluebugs has quit [Read error: Connection reset by peer]
bluebugs has joined #dri-devel
nchery has joined #dri-devel
bluebugs has quit [Ping timeout: 480 seconds]
hanetzer has joined #dri-devel
tursulin has quit [Read error: Connection reset by peer]
<clever>
if glxgears is using radeonsi_dri.so, can i rebuild mesa with just -Dgallium-drivers=radeonsi, and everything will still work? do i need swrast at all?
<airlied>
nope that should be enough
rasterman has quit [Quit: Gettin' stinky!]
<clever>
currently, my distro is using auto, and i wanted to add vc4/v3d, and i might as well ditch the ones i'm not using while i rebuild
rasterman has joined #dri-devel
<clever>
`galliumDrivers = [ "radeonsi" "vc4" "v3d" ];` is my current plan
<mattst88>
you're able to use the same build for raspberrypi and radeonsi?
<mattst88>
what kind of arm system do you have with a radeonsi attached?
<airlied>
mattst88: running the vc4 on shim on x86 I think is the goal
<mattst88>
ah
<anholt_>
mattst88: not nearly the first person I've seen use vc4 on x86. you pipe your emulated x86 wine across to the host to play games.
<mattst88>
oh wow, that's neat
<anholt_>
(I originally learned of this when someone sent me SSE asm for vc4)
<mattst88>
hah!
<clever>
anholt_: in my case, the goal is to pre-compile GLSL to the QPU shader binary, and then embed it into a non-linux app