#
#asahi-gpu · 2021-04-11 02:14
dougallbloom: oh, wow, that's awesome! fun to see how the goal of generating html turned into the possiblity of generating xml/c
#asahi-gpu · 2021-04-11 01:52
DarkShadow44directly or using an IR?
#asahi-gpu · 2021-04-11 01:52
bloomNIR to real code, yes
#asahi-gpu · 2021-04-11 01:48
DarkShadow44bloom: compiler, like in from NIR to bytecode?
#asahi-gpu · 2021-04-11 01:45
bloomhave most of the big ideas sketched in my head, we'll see how long it takes me to get to simple shaders
#asahi-gpu · 2021-04-11 01:45
bloomI started writing the compiler
#asahi-gpu · 2021-04-11 01:45
bloomAnyway uhm
#asahi-gpu · 2021-04-11 01:36
bloomI guess
#asahi-gpu · 2021-04-11 01:29
DarkShadow44but making something useful in the process would be nice too :P
#asahi-gpu · 2021-04-11 01:28
DarkShadow44heh, sure
#asahi-gpu · 2021-04-11 01:27
bloomI thought we were just having fun :)
#asahi-gpu · 2021-04-11 01:27
bloomDuplicating effort?
#asahi-gpu · 2021-04-11 01:26
DarkShadow44would be good to have feedback before I implment too much though, don't wanna duplicate the effort
#asahi-gpu · 2021-04-11 01:23
bloomOkey dokey
#asahi-gpu · 2021-04-11 01:22
DarkShadow44Because I still have my propsed architecture from my asmtool
#asahi-gpu · 2021-04-11 01:21
DarkShadow44bloom: okay then
#asahi-gpu · 2021-04-11 01:16
bloomNothing is set in stone yet
#asahi-gpu · 2021-04-11 01:16
bloomstill experimenting
#asahi-gpu · 2021-04-11 01:16
bloomTBD
#asahi-gpu · 2021-04-11 00:27
DarkShadow44bloom: Even for assemling shaders?
#asahi-gpu · 2021-04-10 23:13
bloom( regardless https://gitlab.freedesktop.org/asahi/applegpu/-/blob/xml/agx_disasm.c.py and https://gitlab.freedesktop.org/asahi/applegpu/-/blob/xml/dump.py is today's yak shave )
#asahi-gpu · 2021-04-10 23:13
bloomThough tbh I am thinking we'd be happier just pulling applegpu.py into mesa
#asahi-gpu · 2021-04-10 23:11
bloomhas a WIP XMLification of applegpu's disassembler
#asahi-gpu · 2021-04-10 22:40
dougallDarkShadow44: good catch, thanks - i'll try to look later today, but i'm guessing 0b00 is the correct encoding there https://github.com/dougallj/applegpu/blob/main/applegpu.py#L859-L860
#asahi-gpu · 2021-04-10 18:26
DarkShadow44Would you mind taking a look at this: https://pastebin.com/0f6ey9vu
#asahi-gpu · 2021-04-10 18:26
DarkShadow44dougall: I found some issue with your documentation
#asahi-gpu · 2021-04-10 17:54
bloom("Go back to sleep, Bloom.")
#asahi-gpu · 2021-04-10 17:51
bloomAnd for my next trick, I shall pull a rabbit out of a hat!
#asahi-gpu · 2021-04-10 17:06
DarkShadow44doesn't work for computer shader yet?
#asahi-gpu · 2021-04-10 17:06
bloomvert in this case
#asahi-gpu · 2021-04-10 17:06
bloommm
#asahi-gpu · 2021-04-10 17:06
DarkShadow44bloom: Using pandecode_cmdstream after a computer shader?
#asahi-gpu · 2021-04-10 17:02
glibcalright :)
#asahi-gpu · 2021-04-10 17:02
bloomworth checking as well, and would probably be easier :)
#asahi-gpu · 2021-04-10 17:01
glibchow about having a function-local array, dynamically indexed by a uniform/attribute?
#asahi-gpu · 2021-04-10 17:00
bloomDarkShadow44: modifying a shader to vary its register pressure, disassembling the shader to get the amount spilled (X), and parsing the command stream to get the amount reported (Y)
#asahi-gpu · 2021-04-10 16:59
DarkShadow44bloom: Out of curiousity, how does one get those test numbers? I'd be interested in the result of (for example) 800
#asahi-gpu · 2021-04-10 16:45
bloomsomething like upper bits are powers of 4, lower bits are 2/3
#asahi-gpu · 2021-04-10 16:45
bloomjn__: Yeah, I think you got it spot on.
#asahi-gpu · 2021-04-10 16:45
jn__bloom: the pattern i fitted is like 1 - 2 3 4 - 8 12 16 -- one big step (double the number), two small steps (increment the two most significant bits), repeat
#asahi-gpu · 2021-04-10 16:45
bloomit's not about division by 3 at all, gah
#asahi-gpu · 2021-04-10 16:44
bloomd'oh
#asahi-gpu · 2021-04-10 16:44
bloomjn__: ty
#asahi-gpu · 2021-04-10 16:44
bloom----Oh shit
#asahi-gpu · 2021-04-10 16:43
jn__finds this table much easier to look at in hex
#asahi-gpu · 2021-04-10 16:43
bloomBut again... why 3?!
#asahi-gpu · 2021-04-10 16:43
bloomglibc: The "explanation" there is if you divide by 3, that becomes a 124/129 boundary -- fine, 128
#asahi-gpu · 2021-04-10 16:42
bloomme too
#asahi-gpu · 2021-04-10 16:42
bloomglibc: yeah
#asahi-gpu · 2021-04-10 16:42
bloom(388, 7)

plus ›