marcan changed the topic of #asahi-gpu to: Asahi Linux: porting Linux to Apple Silicon macs | GPU / 3D graphics stack black-box RE and development (NO binary reversing) | Keep things on topic | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-gpu
bpye has quit [Ping timeout: 240 seconds]
bpye has joined #asahi-gpu
taowa has quit [Ping timeout: 245 seconds]
brinly has quit [Ping timeout: 258 seconds]
jsvcycling_ has quit [Ping timeout: 258 seconds]
jsvcycling_ has joined #asahi-gpu
brinly has joined #asahi-gpu
taowa has joined #asahi-gpu
phiologe has quit [Ping timeout: 250 seconds]
phiologe has joined #asahi-gpu
zkrx has quit [Ping timeout: 240 seconds]
zkrx has joined #asahi-gpu
odmir has quit [Remote host closed the connection]
bpye has quit [Ping timeout: 240 seconds]
bpye has joined #asahi-gpu
mxw39 has quit [Quit: Konversation terminated!]
mxw39 has joined #asahi-gpu
mxw39 has quit [Remote host closed the connection]
mxw39 has joined #asahi-gpu
mxw39 has quit [Remote host closed the connection]
mxw39 has joined #asahi-gpu
mxw39 has quit [Remote host closed the connection]
mxw39 has joined #asahi-gpu
mxw39 has quit [Quit: Konversation terminated!]
mxw39 has joined #asahi-gpu
modwizcode has quit [Remote host closed the connection]
mxw39 has quit [Quit: Konversation terminated!]
mxw39 has joined #asahi-gpu
mxw39 has quit [Remote host closed the connection]
mxw39 has joined #asahi-gpu
mxw39 has quit [Remote host closed the connection]
mxw39 has joined #asahi-gpu
mxw39 has quit [Client Quit]
mxw39 has joined #asahi-gpu
Necrosporus has quit [Ping timeout: 240 seconds]
odmir has joined #asahi-gpu
odmir has quit [Ping timeout: 240 seconds]
Necrosporus has joined #asahi-gpu
odmir has joined #asahi-gpu
odmir has quit [Ping timeout: 240 seconds]
odmir has joined #asahi-gpu
odmir has quit [Remote host closed the connection]
odmir has joined #asahi-gpu
odmir has quit [Remote host closed the connection]
odmir has joined #asahi-gpu
odmir has quit [Ping timeout: 260 seconds]
odmir has joined #asahi-gpu
odmir has quit [Ping timeout: 240 seconds]
odmir has joined #asahi-gpu
_whitelogger has joined #asahi-gpu
titanous has left #asahi-gpu [#asahi-gpu]
vlixa has quit [Remote host closed the connection]
vlixa has joined #asahi-gpu
<bloom> Let's eee
<bloom> Let's eee
* jn__ eees
* bloom is baffled how AGX command buffers are supposed to work
<bloom> There's... not nearly as much structure as I would expect
<chrisf> bloom: something we can look at?
<bloom> chrisf: Just prodding demo.c
<bloom> Specifically what all the magic numbers in demo_cmdbuf could be
<bloom> are these commands? data structures? a mix?
<bloom> after chasing its pointer to unk2, things around unk2 make a bit more sense
<bloom> also, the unaligned pointer thing is bothering me
<chrisf> yeah, that's weird
<bloom> ....are these register writes?
<bloom> still super weird.
<chrisf> looking at how the clear works-- seems a lot of machinery to avoid passing the clear color in the proper format to start with?
<bloom> What clear format?
<bloom> *proper
<bloom> The clear is just a shader, apparently. Because of course it is.
<chrisf> there's a preshader that converts it to fp16?
<bloom> Yes.
<bloom> let me splice that out of the demo and do it on the CPU since you're right, it's ridiculous
<bloom> chrisf: Also, do you know if there's a standard name for preshaders?
<chrisf> i havent heard much chatter about them for a long time. in the old dx9 days they were preshaders
<bloom> I just mean 'compute kernel running right before to precompute a dynamically uniform value"
<chrisf> mali does (/ 'has to do') some amount of this, right?
<bloom> all the proprietary drivers use them, AGX just is a bit extra about it
<bloom> mali does but not so aggressively since there's overhead to it
<chrisf> afaik intel has never bothered to
<chrisf> i imagine mostly because graphics <-> compute switch was until very recently brutal
<chrisf> "now we will reconfigure the universe... "
<bloom> yeah
<bloom> whereas AGX makes it ~free
<chrisf> afaik the open drivers generally dont because https://patents.google.com/patent/US20080291207A1/en
<bloom> what's that?
<bloom> chrisf: ok, updated the demo to do the conversion CPU side which is both simpler and faster :)
<chrisf> neat
<bloom> Then again I don't understand why clears need a shader at all