01:17
<
bloom >
dougall: Passed my first test with loops
01:17
<
bloom >
thanks for the delicious notes
01:18
<
PthariensFlame[m] >
Yay!
01:19
<
bloom >
more work remains but hey
02:04
<
dougall >
:o well done!
02:12
koorogi has joined #asahi-gpu
02:28
phiologe has joined #asahi-gpu
02:29
<
bloom >
dougall: You did the hard part I just typed it in :p
02:32
PhilippvK has quit [Ping timeout: 480 seconds]
02:47
aquijoule__ has joined #asahi-gpu
02:51
<
bloom >
Passed: 625/625 (100.0%)
02:52
<
bloom >
dougall: ^ dEQP-GLES2.functional.shaders.loops.*
02:53
aquijoule_ has quit [Ping timeout: 480 seconds]
02:54
<
dougall >
hahaha that's awesome! does that imply most/all control flow is working, or do the loop tests avoid mixing other structures?
02:54
<
bloom >
dougall: it's everything :)
02:54
<
bloom >
`discard` is separate
02:55
<
bloom >
I haven't implemented discard yet since they don't have a discard_if which meant having if-stmts working was a prereq
02:55
<
bloom >
also it has some weird interactions with TODO.writeout
02:55
<
bloom >
(Do we know what TODO.writeout actually does?)
02:58
<
dougall >
(not in any depth - i think roughly setting up the cf state for interpolation - but i think you named it and probably understand better)
03:01
<
bloom >
~~i thought you named it~~
03:01
<
bloom >
instruction hot potato
03:01
<
bloom >
uhm, inteprolation is unlikely
03:01
<
bloom >
since it executes /after/ the ld_vary instruction in usual circumstnaces
03:01
<
bloom >
it definitely related to writeout, in that I know /where/ it executes
03:02
<
bloom >
in a simple frag shader, right before `blend`
03:02
<
bloom >
er, two before blend
03:02
<
bloom >
in a frag shader with multiple outputs (render targets), two before the frist blend and one before each subsequent blend
03:02
<
bloom >
but then it gets trickier
03:02
<
bloom >
if fragment depth/stencil/coverage is written, that interacts
03:02
<
dougall >
oh, i'm getting mixed up i guess - i thought it was at the end of the vertex shaders?
03:03
<
bloom >
nah, end of frag
03:03
<
bloom >
if discard is used, that interacts (keep in mind `discard` is just a glorified `coverage mask = 0` write!)
03:03
<
bloom >
errrr except Metal's discard is actually demote()
03:05
<
bloom >
yeah, spec says prior to Metal 2.3 it was discard(), Metal 2.3+ it's demote()
03:12
<
dougall >
yeah - having given it some thought, i definitely have no idea :P
03:14
<
dougall >
(i think i was doing offscreen rasterisation tests, and you don't get a writeout instruction there? it just stores to tile memory?)
03:15
<
bloom >
i'm almost certain you get them offscreen too
03:15
<
bloom >
`writeout` is probably a misnomer though
03:15
<
bloom >
but it's used in conjunction with `blend` to write to the tile buffer, i guess
03:15
<
bloom >
quite possibly it has to do with ordering, or something
03:16
<
dougall >
hmm, yeah, i'm probably doing other weird things that are confusing me
03:19
<
bloom >
there are lots of funny interactions
03:28
<
bloom >
Pass: 12351, Fail: 1385, Crash: 366, Warn: 27, Skip: 89, Timeout: 12, Flake: 36, Duration: 10:02, Remaining: 0
03:32
<
bloom >
so 86.7%, up 2% from May 8 when i last ran deqp
03:34
<
bloom >
though tbh anything <99.5% is unusable :P
06:53
keenriser has joined #asahi-gpu
06:59
EER has quit [Ping timeout: 480 seconds]
08:12
tardyp has quit [Read error: Connection reset by peer]
08:20
jabashque has quit [Remote host closed the connection]
08:26
jabashque has joined #asahi-gpu
08:52
choozy has joined #asahi-gpu
09:03
_whitelogger has joined #asahi-gpu
09:06
_whitelogger has joined #asahi-gpu
09:13
_whitelogger has joined #asahi-gpu
09:15
_whitelogger has joined #asahi-gpu
09:19
_whitelogger has joined #asahi-gpu
10:20
choozy has quit [Ping timeout: 480 seconds]
10:35
Simonx22 has joined #asahi-gpu
12:51
ave has joined #asahi-gpu
13:01
choozy has joined #asahi-gpu
14:08
radex has joined #asahi-gpu
14:35
choozy has joined #asahi-gpu
14:54
_whitelogger has joined #asahi-gpu
18:12
keenriser has quit [Remote host closed the connection]
20:06
kubes has joined #asahi-gpu
20:25
dottedmag has quit [Quit: QUIT]
20:26
dottedmag has joined #asahi-gpu