00:25
Dcow has joined #asahi-gpu
00:33
Dcow has quit [Ping timeout: 480 seconds]
00:43
swaggie has quit [Remote host closed the connection]
00:44
swaggie has joined #asahi-gpu
00:53
swaggie has quit [Ping timeout: 480 seconds]
00:59
Dcow has joined #asahi-gpu
01:07
Dcow has quit [Ping timeout: 480 seconds]
01:34
Dcow has joined #asahi-gpu
01:34
Gues__________________________ has joined #asahi-gpu
01:42
Dcow has quit [Ping timeout: 480 seconds]
01:42
Gues__________________________ has quit [Ping timeout: 480 seconds]
01:48
<
alyssa >
Looking into rasterizer discard + vertex shader side effects
01:49
<
alyssa >
which will be convenient for transform feedback, although there's no great way to expose in GL
01:49
<
alyssa >
I have a simple store shader, it looks like what's expected..
01:49
<
alyssa >
except for a magic "TODO.unk51"
01:50
<
alyssa >
in between the store and the stop
01:50
cylm_ has quit [Ping timeout: 480 seconds]
01:50
<
alyssa >
There are 4 outputs identified, oddly
01:50
<
alyssa >
output select is false all the way through, identifying them as the position trivially.
01:50
<
alyssa >
bitmask 0x30 of VDM State Vertex Unknown is now set
01:51
<
alyssa >
no colour output is enabled in the frag control 2
01:51
<
alyssa >
(like a depth only pass)
01:51
<
alyssa >
new bit in Cull 2 enabled
01:52
<
alyssa >
no fragment shader, unsure how that's unbound here
01:52
<
alyssa >
oh, and bit 8 of word 99 of IOGPU Graphics is set
01:53
<
alyssa >
iff there is a store in the VS
01:53
<
alyssa >
for an empty VS with rast discard, the program has the unk51 at the end
01:53
<
alyssa >
Possibly this is a trivial "done with UVS" instruction
01:53
<
alyssa >
whereas usually that would be signalled with the "st_vary_final"
01:56
<
alyssa >
certainly it's encoded like st_var
01:57
<
alyssa >
got the same upper bit encoded
02:09
Dcow has joined #asahi-gpu
02:10
systwi has quit [Ping timeout: 480 seconds]
02:11
systwi has joined #asahi-gpu
02:14
<
alyssa >
trying to replicate on the linux side, so far i'm getting faults.
02:15
<
alyssa >
hm i wonder if occlusion query 2 matters, that's different too
02:17
Dcow has quit [Ping timeout: 480 seconds]
02:17
<
alyssa >
With the magic "no varyings" instruction inserted it doesn't fault
02:18
<
alyssa >
and maybe? works?
02:26
<
alyssa >
I think the "stores from VS" functionality works now
02:26
<
alyssa >
and I'm back to plumbing in streamout
02:39
<
alyssa >
Passed my first XFB test
02:46
Dcow has joined #asahi-gpu
02:58
Dcow has quit [Ping timeout: 480 seconds]
02:59
bpye8 has joined #asahi-gpu
03:01
<
alyssa >
Passed: 1212/1320 (91.8%)
03:02
bpye has quit [Ping timeout: 480 seconds]
03:02
bpye8 is now known as bpye
03:24
<
alyssa >
that's all of them (-:
03:43
Dcow has joined #asahi-gpu
03:45
<
alyssa >
(Is it correct? Well, no, but it is transform feedback.)
03:51
digicyc has quit [Remote host closed the connection]
03:51
Dcow has quit [Ping timeout: 480 seconds]
03:51
digicyc has joined #asahi-gpu
03:55
Dcow has joined #asahi-gpu
03:59
<
alyssa >
For dEQP-GLES3:
03:59
<
alyssa >
Pass: 43858, Fail: 42, Crash: 26, Warn: 37, Skip: 209, Duration: 8:07, Remaining: 0
03:59
<
alyssa >
99.8% passing
04:03
Dcow has quit [Ping timeout: 480 seconds]
04:24
capta1nt0ad has joined #asahi-gpu
04:28
Dcow has joined #asahi-gpu
04:39
Dcow has quit [Ping timeout: 480 seconds]
04:43
capta1nt0ad has quit [Read error: Connection reset by peer]
05:00
c10l has quit [Quit: Bye o/]
05:01
c10l has joined #asahi-gpu
05:06
<
alyssa >
hm I have both good news and bad news.
05:06
<
alyssa >
oh wait duh
05:08
maria7 has joined #asahi-gpu
05:11
maria has quit [Ping timeout: 480 seconds]
05:11
maria7 is now known as maria
05:13
handlerug has quit [Read error: No route to host]
05:13
ella-0 has quit [Read error: No route to host]
05:13
handlerug has joined #asahi-gpu
05:13
ella-0 has joined #asahi-gpu
05:23
Dcow has joined #asahi-gpu
05:31
Dcow has quit [Ping timeout: 480 seconds]
06:06
Dcow has joined #asahi-gpu
06:14
Dcow has quit [Ping timeout: 480 seconds]
06:24
swaggie has joined #asahi-gpu
06:40
kode54 has quit [Quit: Ping timeout (120 seconds)]
06:41
kode54 has joined #asahi-gpu
06:44
SSJ_GZ has joined #asahi-gpu
07:49
Dcow has joined #asahi-gpu
07:59
Dcow has quit [Ping timeout: 480 seconds]
09:05
Dcow has joined #asahi-gpu
09:13
Dcow has quit [Ping timeout: 480 seconds]
09:42
Dcow has joined #asahi-gpu
09:48
iaguis has joined #asahi-gpu
09:50
Dcow has quit [Ping timeout: 480 seconds]
10:14
MajorBiscuit has joined #asahi-gpu
10:20
Dcow has joined #asahi-gpu
10:35
Dcow has quit [Ping timeout: 480 seconds]
10:42
bcrumb has joined #asahi-gpu
10:47
bcrumb has quit [Quit: WeeChat 3.7.1]
10:47
bcrumb has joined #asahi-gpu
10:59
swaggie has quit [Remote host closed the connection]
11:00
swaggie has joined #asahi-gpu
11:09
swaggie has quit [Ping timeout: 480 seconds]
11:18
Dcow has joined #asahi-gpu
11:23
bcrumb has quit [Quit: WeeChat 3.7.1]
11:26
Dcow has quit [Ping timeout: 480 seconds]
11:34
iaguis has quit [Ping timeout: 480 seconds]
11:40
swaggie has joined #asahi-gpu
11:52
Dcow has joined #asahi-gpu
12:00
Dcow has quit [Ping timeout: 480 seconds]
12:01
Dcow has joined #asahi-gpu
12:06
<
jannau >
I never tested what happens what happens when surface[3] is used. I wasn't expected that it is simply ignored
12:10
Dcow has quit [Ping timeout: 480 seconds]
12:12
swaggie has quit [Ping timeout: 480 seconds]
12:36
Dcow has joined #asahi-gpu
12:47
Dcow has quit [Ping timeout: 480 seconds]
12:51
Hibyehello has joined #asahi-gpu
13:06
swaggie has joined #asahi-gpu
13:14
swaggie has quit [Ping timeout: 480 seconds]
14:29
Dcow has joined #asahi-gpu
14:38
Dcow has quit [Ping timeout: 480 seconds]
15:05
Dcow has joined #asahi-gpu
15:13
Dcow has quit [Ping timeout: 480 seconds]
15:15
swaggie has joined #asahi-gpu
15:17
iaguis has joined #asahi-gpu
15:19
le0n has quit [Quit: see you later, alligator]
15:20
le0n has joined #asahi-gpu
15:21
iaguis_ has joined #asahi-gpu
15:23
swaggie has quit [Ping timeout: 480 seconds]
15:24
Dcow has joined #asahi-gpu
15:32
Dcow has quit [Ping timeout: 480 seconds]
15:33
iaguis has quit [Quit: leaving]
15:33
iaguis_ has quit [Quit: leaving]
15:33
iaguis has joined #asahi-gpu
15:42
alyssa has quit [Quit: leaving]
16:11
pb17 has joined #asahi-gpu
16:17
DarkShadow44 has joined #asahi-gpu
16:20
DarkShadow4444 has joined #asahi-gpu
16:25
DarkShadow44 has quit [Ping timeout: 480 seconds]
16:31
cylm_ has joined #asahi-gpu
16:46
le0n_ has joined #asahi-gpu
16:51
yuka_ has quit [Remote host closed the connection]
16:53
le0n has quit [Ping timeout: 480 seconds]
17:04
MajorBiscuit has quit [Ping timeout: 480 seconds]
17:16
swaggie has joined #asahi-gpu
17:21
yuka has joined #asahi-gpu
17:24
swaggie has quit [Ping timeout: 480 seconds]
17:39
<
jannau >
marcan: matches my ovservations. the 384..896 range had me confused. native primaries are probably display P3
17:40
MajorBiscuit has joined #asahi-gpu
17:45
bcrumb has joined #asahi-gpu
17:48
<
marcan >
jannau: updated with 8-bit mode observations
17:53
<
marcan >
I think 9 is BT.2020 and 12 is just native
17:54
<
marcan >
jannau: I'm not sure native is actually P3. I think internally they use P3 (the rtings.com shows the display as almost perfectly matching P3, which is... unlikely) but the display can do more
17:54
<
marcan >
I see significant red clipping in BT.2020 mode, but only a bit of green clipping. if it were strictly P3 I'd expect more green clipping
17:55
<
marcan >
so I think the green primary is purer than P3
17:56
<
marcan >
the colorimetry table we have says 9 is BT.2020 and 12 is "scRGBfixed" which sounds like a plausible code for "native whatever" or something abused as such
17:57
<
marcan >
this also means we should probably be using colorimetry 12 for both 8-bit and 10-bit modes for now, as that gives full uncalibrated range (oversaturated for sRGB, but hey people like that)
17:57
<
marcan >
and then we should implement the proper colorimetry connector modes
17:58
<
marcan >
however, the weirdo range bothers me. if it turns out it's that DisplayPort thing, Linux already actually has something for that
17:58
<
marcan >
if not though, I think we're going to have to use the matrix stuff to expand it back to 0..1024 for most software to not be confused (or at least offer that option)
18:10
swaggie has joined #asahi-gpu
18:23
alyssa has joined #asahi-gpu
18:23
<
alyssa >
Pass: 43858, Fail: 34, Crash: 26, Warn: 37, Skip: 209, Missing: 225, Duration: 8:16, Remaining: 0
18:23
<
alyssa >
okay yeah that's more like it
18:23
<
alyssa >
wait why 225 missing
18:26
<
alyssa >
whoops. let's try again
18:34
zkrx has joined #asahi-gpu
18:37
<
alyssa >
Pass: 44083, Fail: 34, Crash: 26, Warn: 37, Skip: 209, Duration: 8:05, Remaining: 0
18:37
<
alyssa >
Ok, that's better
18:38
<
alyssa >
depending how you count, 99.8% or 99.9%
18:39
<
alyssa >
mostly multisampling but some other stuff too
18:39
<
alyssa >
bunch of cube map fails, that's concerning..
18:40
<
alyssa >
nearest filtering with NPOT cubemaps I guess
18:44
<
alyssa >
Nothing glaringly wrong
19:25
<
alyssa >
piglit is... less impressive
19:25
<
alyssa >
Pass: 13356, Fail: 2219, Crash: 493, Skip: 4625, Timeout: 4, Flake: 2, Duration: 4:20, Remaining: 0
19:29
bcrumb has joined #asahi-gpu
19:33
<
alyssa >
Well, that's why agx/next isn't released :p
19:39
<
alyssa >
Why is it allowed to mipmap 1D textures? well.
20:07
<
alyssa >
ext_transform_feedback-immediate-reuse needs barriers between the vertex jobs, fun
20:36
iaguis has quit [Ping timeout: 480 seconds]
20:36
bcrumb has quit [Quit: WeeChat 3.7.1]
20:49
alyssa has quit [Quit: leaving]
21:09
iaguis has joined #asahi-gpu
21:52
iaguis has quit [Quit: leaving]
22:03
swaggie has quit [Remote host closed the connection]
22:40
bcrumb has joined #asahi-gpu
22:58
bcrumb has quit [Quit: WeeChat 3.7.1]
23:10
bcrumb has joined #asahi-gpu
23:23
bpye has quit [Quit: Ping timeout (120 seconds)]
23:23
bpye has joined #asahi-gpu
23:28
MajorBiscuit has quit [Ping timeout: 480 seconds]
23:32
SSJ_GZ has quit [Ping timeout: 480 seconds]