marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | General project discussion | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
skali has quit [Ping timeout: 480 seconds]
grange_c has quit [Quit: Ping timeout (120 seconds)]
grange_c has joined #asahi
ella-0 has quit [Ping timeout: 480 seconds]
<milek7>
does M1 have accelerated opengl on macos?
<milek7>
because I have report of application running so slow that I think it is falls back to software rendering
<i509vcb[m]>
I believe it's OpenGL implemented ontop of Metal
<i509vcb[m]>
Per vendor info in Minecraft
<i509vcb[m]>
Unless that's a symptom of software renderer?
<alyssa>
milek7: Yes.
<alyssa>
i509vcb[m]: This is correct, it's GL-on-Metal, not software.
skali has joined #asahi
Tom__ has quit [Read error: Connection reset by peer]
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
malvo has quit [Ping timeout: 480 seconds]
malvo has joined #asahi
skali has joined #asahi
PhilippvK has joined #asahi
phiologe has quit [Ping timeout: 480 seconds]
skali has quit [Ping timeout: 480 seconds]
borrrden has joined #asahi
marvin24 has joined #asahi
marvin24_ has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
yuyichao_ has joined #asahi
borrrden has quit [Quit: WeeChat 2.8]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
skali has quit [Ping timeout: 480 seconds]
skali has joined #asahi
<marcan>
maz: I know you expressed some interest in the cpufreq/idle stuff, are you looking into that? otherwise I was thinking of starting to poke around that, build some m1n1 tests to measure relative perf and switching latencies
<marcan>
well, I'll do the tests anyway, but let me know if you want to poke at the kernel side after that :)
<marcan>
I had a look at how some other socs represent the cluster-level perf states and such
<marcan>
also, if anyone has shell.py handy, I'd like to know if you get any different values for this:
<marcan>
that should be related to memory controller performance; the ADT describes a higher performance state and the driver apparently just uses the current state as a default, but if that is always the same I might want to just have both described
<sven>
same values on my mac mini
<marcan>
the perf value is hardcoded in the master adt too, not substituted in, so I think we can safely hardcode these
<marcan>
just pushed proper support for template ADTs in m1n1.adt
<marcan>
there's actually relatively little that is per-device in here
<marcan>
at least little that is templated like this
<marcan>
other than product info and MACs, there's:
<marcan>
yeah, rosaline is the flood illuminator AIUI... what's this doing on the MBP, that makes no sense
<marcan>
this is under isp {}, I think it's just junk left over from iPhones
bgb has joined #asahi
<Ariadne>
this is really impressive to see how far asahi has come in such a short time
<Ariadne>
i think the mac mini, with asahi's work, is going to be the preferred ARM workstation solution
<Ariadne>
the "competition" is proposing a design with a price tag around $2000
<ar>
🤷 for me the current m1 lineup is undesirable because of its 16G ram limit
<ar>
even the lx2k board i have has 2 sodimm slots with 32GB of ram
<sven>
i'm hoping for a m1x/m2/whatever with 32gb ram
<marcan>
pretty sure that's coming
<sven>
i'd be surprised if it isn't :)
<marcan>
in fact the only reason I din't bui 3-4 minis to replace my home server with an actual redundant build/storage cluster is because I'm waiting for the M1X ;)
<marcan>
*buy
<Emantor[m]>
I doubt that the layerscape 2160 on lx2k can match the performance of the M1. Lots of ethernet throughput though, which is nice :-)
bgb has quit [Ping timeout: 480 seconds]
<ar>
Emantor[m]: single thread, absolutely not
<kettenis>
marcan: not sure what the wifi calibration stuff is for
<_jannau_>
16 core lx2k was iirc factor 4 slower than 8 core m1 (running linux in a VM) NEON heavy workload (AV1 decoding)
<kettenis>
don't think the current brcmfmac driver has a way to use that stuff
<marcan>
I also see iBoot making quite a few changes outside of those templates
<marcan>
adding tunables etc
<marcan>
sven: what version are you on, and would you mind sending me your DT? I wonder how much actually changes between units
<marcan>
*ADT
<marcan>
(warning: likely has your wifi password in it)
<sven>
no idea what version i'm on
<sven>
not sure i even setup wifi, but let me check
<sven>
huh... it has my wifi stuff in there
<marcan>
you can look at system-firmware-version and firmware-version
<sven>
oh... icloud sync probably
<sven>
firmware-version: iBoot-6723.81.1
<sven>
system-firmware-version: iBoot-6723.81.1
<marcan>
ah, older
<marcan>
might still be good enough for diffing tunables though
<Emantor>
system-firmware-version = iBoot-6723.120.36, firmware-version = b'iBoot-6723.120.36 for my mini.
<sven>
yeah, let me just strip the wifi stuff
<marcan>
going to stream for a bit, looking at the cpufreq stuff
<Emantor>
Looks like one of the macOS updates pulled in a newer version and i was looking at an older dump, system-firmware-version = iBoot-6723.140.2 & firmware-version = iBoot-6723.120.36 is whats currently on my m1n1.
skali has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
skali has joined #asahi
skali is now known as skali_
skali_ is now known as skali
skali has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
X-Scale` has joined #asahi
X-Scale has quit [Ping timeout: 480 seconds]
X-Scale has joined #asahi
X-Scale` has quit [Ping timeout: 480 seconds]
al3xtjames8 has joined #asahi
al3xtjames has quit [Ping timeout: 480 seconds]
al3xtjames8 is now known as al3xtjames
chadmed has quit [Remote host closed the connection]
chadmed has joined #asahi
aleasto has joined #asahi
nskl has quit [Ping timeout: 480 seconds]
yuyichao_ has quit [Ping timeout: 480 seconds]
psykose has quit [Ping timeout: 480 seconds]
psykose has joined #asahi
nskl has joined #asahi
<marcan>
lol
psykose has quit [Ping timeout: 480 seconds]
<marcan>
I know why I couldn't boost to 3.2
<marcan>
macos has the same issue
<marcan>
I just tested it on the MBA: the cores claim to be at 3.2, but if all are active, the cluster gets capped at 3.0
<marcan>
so you need to actually put some cores to sleep to reach the higher pstates
<marcan>
well that was a waste of time on the stream :p
<j_ey>
*actual* sleep?
<j_ey>
like wfi or whatever, or just lower states
<marcan>
the state is per-cluster
<marcan>
cores can only be in sleep states
<marcan>
which is two kinds of wfi, the kind that loses state and the kind that does not
<j_ey>
oh right, state is per cluster
<marcan>
well, either that or it's a thermal issue
<marcan>
the MBA extremely quickly throttles down from max state to the prior one
<marcan>
but I think it's not that
<kettenis>
the mini should be fine; it has a fan
<rowang077[m]>
Do you develop this on an air? I thought you had a mac mini.
yuyichao_ has joined #asahi
<j_ey>
rowang077[m]: marcan has all of them!
psykose has joined #asahi
<marcan>
== Final CPU frequencies ==
<marcan>
CPU 0: 599.99 MHz
<marcan>
CPU 4: 3203.84 MHz
<marcan>
there.
<j_ey>
marcan: please feed yourself now!
<marcan>
switched the SMP code to use wfi and IPIs to wake up secondaries, and made a deep sleep function
<marcan>
also only a single register write is needed
<marcan>
everything I did on the stream was a waste of time :)
<j_ey>
lol ;_;
<marcan>
(well, I did want to bring up PMP anyway for power measurements, eventually)
zigmars has joined #asahi
zigmars has quit [Remote host closed the connection]
zigmars has joined #asahi
<marcan>
(pushed)
zigmars has quit [Read error: Connection timed out]
zigmars has joined #asahi
zigmars has quit [Remote host closed the connection]
zigmars has joined #asahi
Andalu30 has joined #asahi
Andalu30 has quit [Ping timeout: 480 seconds]
Andalu30 has joined #asahi
jvoisin has quit [Ping timeout: 480 seconds]
zigmars has quit []
jvoisin has joined #asahi
yuyichao_ has quit [Quit: Konversation terminated!]
yuyichao has joined #asahi
<kettenis>
started upstreaming my u-boot code
<j_ey>
kettenis: great!
<alyssa>
woot!
<alyssa>
has anyone else tried kettenis's u-boot with linux?
<alyssa>
I was having SMP and framebuffer problems
<kettenis>
I may have fixed a framebuffer problem today (if you booted using EFI)
<kettenis>
that fix would probably only fix color issues though
<sven>
nice :)
nskl has quit [Ping timeout: 480 seconds]
<kettenis>
sven: The cover letter of the v2 mailbox series said you added some more examples, but I didn't spot those...
<sven>
oh
<sven>
i meant that i wrote into the description that e.g. m3 one is used for thunderbolt
<kettenis>
ah
<sven>
i can add more examples if you think that's helpful though
<kettenis>
so i just misunderstood what you meant
<kettenis>
don't really think that would be particularly helpful
<sven>
hrm, now that i re-read that latter again it's slightly confusing
<sven>
*letter
<alyssa>
letter latter letter latterletter latterletter latter
<psykose>
nerd
<sven>
looks like macos does a full xhci reset whenever a new device is plugged in fwiw
<kettenis>
ouch
<alyssa>
<-- pot
<alyssa>
psykose: <-- kettle
<sven>
and it looks like that may be required :(
<psykose>
alyssa: :p
<sven>
just resetting the port in the PHY doesn't always work
<alyssa>
psykose: you are in the irc channel of a linux port. just saying ;-p
<psykose>
a spicy port at that
nskl has joined #asahi
torstenvl has joined #asahi
torstenvl_ has joined #asahi
torstenvl_ has quit []
torstenvl has quit []
torstenvl has joined #asahi
Andalu30 has quit [Quit: Konversation terminated!]
skali has joined #asahi
skali is now known as skali_
skali_ is now known as skali
skali has quit []
malvo has quit [Ping timeout: 480 seconds]
aleasto has quit [Remote host closed the connection]
torstenvl has quit [Remote host closed the connection]
<zigmars>
I understand that marcan (as seen in worklogs) does a `run_guest.py` with extracted .macho file, so alternatively how & where from would I extract that?
zigmars_ has quit []
<j_ey>
zigmars: if you boot into macOS, you can find a file called `kernelcache` under /System/Volumes/Preboot/
torstenvl has quit [Remote host closed the connection]
zigmars has quit [Remote host closed the connection]
yuyichao has quit [Quit: Konversation terminated!]
zigmars has joined #asahi
<alyssa>
speaking of
<alyssa>
someone who has successfully setup m1n1 and booted linux with usb support and has a flash drive to spare, want to test something for me? for freee!!! 😋
<alyssa>
(I mean. I can provide the kernel image I guess.)
zigmars has quit [Ping timeout: 480 seconds]
zigmars has joined #asahi
zigmars has quit [Remote host closed the connection]
zigmars has joined #asahi
torstenvl has joined #asahi
aleasto has joined #asahi
zigmars has quit []
torstenvl has quit [Ping timeout: 480 seconds]
aleasto has quit [Quit: Konversation terminated!]
artemist has quit [Remote host closed the connection]
artemist has joined #asahi
zigmars has joined #asahi
<alyssa>
I have a few minutes and very little energy. That sounds like a perfect time to rebase my kernel.
<psykose>
rebasing kernels is a health hazard
<alyssa>
shrug
zigmars has quit [Quit: Leaving]
<alyssa>
maz: just built your branch (+ reboot + DCP patches), everything works as before
<alyssa>
in particular hello from m1 ethernet and usb keyboard and mouse :]
<alyssa>
so the pcie series is Tested-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>