00:10
jrmuizel has joined #lima
00:57
jrmuizel has quit [Remote host closed the connection]
01:50
Elpaulo has quit [Read error: Connection reset by peer]
01:52
Elpaulo has joined #lima
02:00
jrmuizel has joined #lima
02:23
_whitelogger has joined #lima
02:28
jrmuizel has quit [Remote host closed the connection]
02:31
dddddd has quit [Read error: Connection reset by peer]
02:43
jrmuizel has joined #lima
02:55
jrmuizel has quit [Remote host closed the connection]
05:11
_whitelogger has joined #lima
05:15
Barada has joined #lima
05:38
_whitelogger has joined #lima
07:10
guillaume_g has joined #lima
07:47
<
cwabbott >
anarsoul: yes, I do have some idea of how to do it :)
08:21
Elpaulo1 has joined #lima
08:23
Elpaulo has quit [Ping timeout: 268 seconds]
08:23
Elpaulo1 is now known as Elpaulo
10:12
Elpaulo1 has joined #lima
10:13
Elpaulo has quit [Ping timeout: 245 seconds]
10:13
Elpaulo1 is now known as Elpaulo
10:30
Barada has quit [Quit: Barada]
11:35
guillaume_g has left #lima ["Konversation terminated!"]
11:51
dddddd has joined #lima
12:06
<
MoeIcenowy >
is mesa master tgsi_to_nir routine broken for Lima?
12:06
Barada has joined #lima
12:06
Barada has quit [Client Quit]
12:26
<
MoeIcenowy >
tried to move control flow flatten code from TGSi codepath to the shared codepath between TGSI and NIR
12:26
<
MoeIcenowy >
glamor seems to be working better on Lima
12:42
<
rellla >
MoeIcenowy: If you force tgsi_to_nir, uniform handling is broken, because nir explicitly expects the variable type set.
12:46
<
MoeIcenowy >
rellla: in fact I used to want to force TGSI because of CF flatten code
12:47
<
rellla >
so then you will run into an assert most likely
12:47
<
MoeIcenowy >
however I found that the code doesn't use TGSi feature
12:47
<
MoeIcenowy >
it's only placed after TGSi and NIR branch
12:48
<
MoeIcenowy >
so I can just move it to the front of the branch
12:48
<
MoeIcenowy >
now I get all if's in FS flatten
12:48
<
MoeIcenowy >
and glamor works here
13:00
jrmuizel has joined #lima
13:11
jrmuizel has quit [Remote host closed the connection]
13:13
Barada has joined #lima
13:13
Barada has quit [Client Quit]
14:03
jrmuizel has joined #lima
14:38
afaerber has joined #lima
15:07
jrmuizel has quit [Quit: Leaving...]
15:11
jrmuizel has joined #lima
15:12
hoijui has joined #lima
15:12
jrmuizel_ has joined #lima
15:16
jrmuizel has quit [Ping timeout: 248 seconds]
15:40
Elpaulo has quit [Quit: Elpaulo]
15:42
drod has joined #lima
15:44
jrmuizel_ has quit [Quit: Leaving...]
15:45
jrmuizel has joined #lima
16:18
ninolein has joined #lima
16:19
<
anarsoul >
MoeIcenowy: then send an MR?
16:21
hoijui has quit [Ping timeout: 264 seconds]
16:56
adjtm has quit [Quit: Leaving]
17:58
drod has quit [Ping timeout: 248 seconds]
18:10
drod has joined #lima
18:57
drod has quit [Remote host closed the connection]
19:57
drod has joined #lima
20:18
ninolein has quit [Ping timeout: 264 seconds]
20:20
ninolein has joined #lima
20:44
hoijui has joined #lima
20:50
drod has quit [Read error: Connection reset by peer]
21:03
drod has joined #lima
21:07
hoijui has quit [Ping timeout: 264 seconds]
21:22
jernej has joined #lima
21:27
afaerber has quit [Remote host closed the connection]
21:34
jrmuizel has quit [Remote host closed the connection]
21:47
jrmuizel has joined #lima
22:22
jrmuizel has quit [Remote host closed the connection]
22:25
afaerber has joined #lima
23:46
<
MoeIcenowy >
anarsoul: finally we still need proper CF support
23:46
<
anarsoul >
MoeIcenowy: I know
23:46
<
MoeIcenowy >
and I don't think there's any driver w/ NIR that have no real CF
23:46
<
MoeIcenowy >
so the flatten looks not so necessary
23:47
<
MoeIcenowy >
(I think the latest known GPU supported by Mesa that has no real CF is i945
23:48
<
MoeIcenowy >
and flatten complex if's really hurts performance