ChanServ changed the topic of #asahi-dev to: Asahi Linux: porting Linux to Apple Silicon macs | Non-development talk: #asahi | General development | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-dev
cylm has quit [Read error: Connection reset by peer]
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<marcan>
j`ey: fixed those myself
seeeath has joined #asahi-dev
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
Kazyek has joined #asahi-dev
___nick___ has joined #asahi-dev
___nick___ has quit []
___nick___ has joined #asahi-dev
Kazyek has quit [Remote host closed the connection]
<sven>
probably a good start is to look at that macvdmtool and see what those commands are
<sven>
then you can probably implement a small proof of concept inside the tipd driver to unconditionally send some of those commands using tps6598x_exec_cmd to make sure they work
<j`ey>
you need 2 macs for this right?
<sven>
yes
<jannau>
https://www.ti.com/lit/ug/slvubh2b/slvubh2b.pdf is the TRM for another variant of the TI chip which has additional bits described and might be a little closer to the variant apple uses
cupermir has quit [Remote host closed the connection]
<sven>
and then finally probably create some debugfs file or something like that to trigger the commands from userspace
<sven>
jannau: good point. I'm still a bit sad that the USB4 variant's datasheets are hidden behind an NDA :(
chadmed has quit [Remote host closed the connection]
cupermir has joined #asahi-dev
chadmed has joined #asahi-dev
tino has joined #asahi-dev
cupermir is now known as Guest3246
tino is now known as cupermir
Guest3246 has quit [Ping timeout: 480 seconds]
hexafox has quit [Ping timeout: 480 seconds]
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jannau>
what did we use to determine capacity-dmips-mhz? I remember coremark but Documentation/devicetree/bindings/arm/cpu-capacity.txt advises for Dhrystone
hightower2 has quit [Ping timeout: 480 seconds]
<chadmed>
iirc marcan used dhrystone on stream
<chadmed>
i used coremark when doing eas testing but ive been using the dhrystone results and they seem to work fine when pipewire is allowed to nice itself to -11
hightower2 has joined #asahi-dev
hightower3 has joined #asahi-dev
hightower2 has quit [Ping timeout: 480 seconds]
<jannau>
I get different results for coremark and dhrystone. e-core capacity of 710 for dhrystone and 756 for coremark
<chadmed>
my results differed from those again
<handlerug>
is there documentation about how DCP works and how the kernel currently talks to it?
<chadmed>
i got 707 for icestorm
<jannau>
dhrystone value is a little suspicious since it's less than icestorm but the m2 e-cores should have become faster relative to the p-cores
<chadmed>
dhrystone just sucks
<jannau>
handlerug: I fear there is only code. C and python code
<chadmed>
why yes i love benchmarking the compilers ability to vectorise this is definitely what i wanted to benchmark thank you
<chadmed>
coremark should have superseded it a long time ago
<chadmed>
not that coremark is _that_ much better
<sven>
jannau: there's also the kext and m1n1 traces, how could you forget these :p
<chadmed>
given that capacity-dmips-mhz is in essence a dimensionless scaling factor i dont really think it matters what we use to derive it, so long as the values are accurate and representative
<jannau>
at least it doesn't produce a ton of warnings like dhrystone, even when compiled with --std=c89
<chadmed>
all the more reason to ditch dhrystone
<jannau>
sven: there's also the firmware but I suppose binary documentation is even worse than C/python code. it has one advantage though. it is by definition correct
<sven>
true. i was mostly joking fwiw ;)
<chadmed>
i think arms official docs say the same thing. like "yes its called dmips but just use whatever is going to work and make sense"
<jannau>
yes, I'll use the coremark results
bluetail12 has joined #asahi-dev
leitao has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leitao has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leitao has joined #asahi-dev
leitao has quit []
leviz has joined #asahi-dev
leviz has quit []
leitao has joined #asahi-dev
leitao has quit []
leviz___ has joined #asahi-dev
leitao has joined #asahi-dev
seeeath has joined #asahi-dev
<kettenis>
m2 pro mini arrived; guess I have to make my own device tree?
<jannau>
you have to start with getting m1n1 to run. see '#define BRINGUP' in config.h
<jannau>
I would be surprised if the HDMI port produces any output, m1n1 might even crash during display config
leitao has quit [Remote host closed the connection]
leitao has joined #asahi-dev
leitao has quit [Remote host closed the connection]