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
greguu has joined #asahi-dev
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
seeeath has quit [Read error: Connection reset by peer]
seeeath has joined #asahi-dev
balrog has quit [Quit: Bye]
balrog has joined #asahi-dev
seeeath has quit []
dk_ has quit [Read error: Network is unreachable]
dk_ has joined #asahi-dev
akspecs has quit [Read error: Network is unreachable]
akspecs has joined #asahi-dev
seeeath has joined #asahi-dev
axboe has quit [Remote host closed the connection]
axboe has joined #asahi-dev
akspecs has quit [Read error: Network is unreachable]
akspecs has joined #asahi-dev
zef_ has joined #asahi-dev
zef has quit [Ping timeout: 480 seconds]
gladiac has joined #asahi-dev
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
uartman_ has joined #asahi-dev
pjakobsson has quit []
bluetail98 has quit []
seeeath has joined #asahi-dev
chadmed has joined #asahi-dev
<chadmed>
alyssa: as lina said it will happen regardless _unless_ the ecores get pushed to 80% util in which case PELT and EAS go out the window and stuff gets shuffled over to less busy cores by force
<chadmed>
that being said, with the current default behaviour and uclamp enabled for FAIR tasks i see no apparent loss in interactivity on the desktop
<chadmed>
(PREEMPT_DYNAMIC and 1000 Hz timer of course)
<marcan>
the problem is going to be corner cases, not light workloads
<marcan>
anyway, we'll see when the recent driver changes are stable and we get the EAS stuff in etc
seeeath has quit [Ping timeout: 480 seconds]
<chadmed>
this is why userspace is being encouraged to start making use of uclamp
<chadmed>
the scheduler will never, ever, have enough information about those types of loads to work well without some help from userspace
<chadmed>
uclamp goes a long way to fixing that
<marcan>
maybe mesa should start uclamping things behind apps' back?
<marcan>
I'd be all for trying that out
<chadmed>
id be interested to see how one would implement that
<chadmed>
youd have to pick some performance metric and then optimise for the minimum uclamp value (0,1024) that lets $app satisfy that
<marcan>
well, for starters you could use 1024:1024. that'd basically turn it into "race to idle" mode, which isn't a bad idea anyway for desktop workloads.
<marcan>
and then we see how much that hurts power consumption and think about other approaches :p
<marcan>
a smarter approach would be to have some serious scheme hooking into the swapbuffers stuff to measure FPS and infer what the app is trying to achieve, and raise uclamp if deadlines are not being met until they are
<chadmed>
that would basically be the same as using the performance governor
<marcan>
yes, but only for those apps
<marcan>
in particular performance right now *sucks* because the CPUs don't drop down in frequency when idle
<marcan>
here they would
<marcan>
and we have fast switching so doing it often is no big deal
<chadmed>
yeah
<chadmed>
inferring fps seems like a decent route but i wonder if that wouldnt just lead to UCLAMP_MIN ending up as 1024 anyway as apps without vsync will just hog all the performance they can get out of the hardware
<marcan>
I wouldn't be surprised if that uclamp gives us >95% of the boost from pinning/performance on glmark, with much better power consumption
<marcan>
yes, apps without vsync would go to 1024 and should
<marcan>
if this gets implemented in mesa it probably should just detect no sync cases and go straight to that
<chadmed>
yeah
<marcan>
also I think mesa has a quirks list, so it could have custom settings per app
<marcan>
like "we know kwin isn't a CPU hog, let it land on the e cores"
<marcan>
(for compositors you want high *priority* but don't necessarily need high *performance*)
<lina>
People are starting to try to run glmark & friends and are really running into this, and reviewers aren't going to know to pin tasks so... Even if it's just a hack at first, I think we really should do this to avoid confusion.
<lina>
Otherwise we're just going to get Phoronix putting out bad benchmarks again...
<marcan>
I guess that's what we get for being the first big.LITTLE system with a decent GPU :p (Intel's mess doesn't count)
<chadmed>
can anyone quickly get the t600x smc key for gpu power?
<chadmed>
gonna do some testing with uclamp now
<chadmed>
i want to log AP and GPU power
chadmed has quit [Remote host closed the connection]
chadmed has joined #asahi-dev
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
MajorBiscuit has joined #asahi-dev
mkurz has quit [Quit: Leaving]
cylm_ has quit [Ping timeout: 480 seconds]
leitao has joined #asahi-dev
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mkurz has joined #asahi-dev
seeeath has joined #asahi-dev
leitao has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
bisko has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
uartman_ has quit [Ping timeout: 480 seconds]
r0ni has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
leitao has joined #asahi-dev
seeeath has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leitao has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
kesslerd has quit [Quit: Leaving]
seeeath has quit [Read error: Connection timed out]
<chadmed>
tldr is schedutil + uclamp has the best perf/W by a decent margin and performs decently close to the performance governor
seeeath has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
chadmed_ has joined #asahi-dev
chadmed has quit [Read error: Connection reset by peer]
leitao has joined #asahi-dev
leitao has quit [Remote host closed the connection]
beeblebrox has quit [Remote host closed the connection]
SuperED has joined #asahi-dev
SuperED has quit [Remote host closed the connection]
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
leitao has joined #asahi-dev
<povik>
huh, upstream admac has a memory leak
<povik>
and i only found out because i copied the relevant code into the sio driver
<povik>
where the consequences are more dire
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
seeeath has quit [Ping timeout: 480 seconds]
seeeath has joined #asahi-dev
stickytoffee has joined #asahi-dev
gladiac has quit [Quit: k thx bye]
uartman_ has joined #asahi-dev
seeeath has quit [Read error: Connection timed out]
seeeath has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sven>
ugh.. looks like I actually have to figure out how to restart the usb4 nhi if I want hotplug between usb4 and „legacy“ thunderbolt to work *sigh*
<maz>
gah. running the SError-triggering EFI payload in a nested guest results in... an SError... Need to propagate the trap bits all the way down... :-/
beeblebrox has joined #asahi-dev
gladiac has joined #asahi-dev
leitao has joined #asahi-dev
uartman_ has quit [Ping timeout: 480 seconds]
<sven>
huh, looks like I can actually bring up pciec in m1n1 as well
<sven>
lets see if the tunables survive a disabled power domain
<sven>
looks like they do :)
<as400>
sven: you should be blogging about your usb fight. Hehe - it's really cool reading it here :) sorry for off
<sven>
ehh… writing a blog post takes much more effort than venting in here ;)
leitao has quit [Ping timeout: 480 seconds]
hightower4 has joined #asahi-dev
hightower3 has quit [Ping timeout: 480 seconds]
cylm_ has joined #asahi-dev
bluetail98 has joined #asahi-dev
leitao has joined #asahi-dev
hightower4 has quit [Ping timeout: 480 seconds]
kesslerd has joined #asahi-dev
MajorBiscuit has quit [Quit: WeeChat 3.6]
kesslerd has quit [Remote host closed the connection]
kesslerd has joined #asahi-dev
kesslerd has quit [Remote host closed the connection]
kesslerd has joined #asahi-dev
hightower2 has joined #asahi-dev
leitao has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sven>
marcan: did you ever get around to investigate why the normal pcie power domains have to be always-on?
cylm has joined #asahi-dev
cylm_ has quit [Ping timeout: 480 seconds]
bluetail986 has joined #asahi-dev
bluetail98 has quit [Ping timeout: 480 seconds]
cylm_ has joined #asahi-dev
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
cylm has quit [Ping timeout: 480 seconds]
maximbaz has quit [Quit: bye]
maximbaz has joined #asahi-dev
hightower2 has quit [Remote host closed the connection]
hightower2 has joined #asahi-dev
r0ni has quit [Ping timeout: 480 seconds]
seeeath has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
seeeath has joined #asahi-dev
gladiac has quit [Quit: k thx bye]
chadmed_ is now known as chadmed
cylm has joined #asahi-dev
cylm_ has quit [Ping timeout: 480 seconds]
darkapex has quit [Remote host closed the connection]
darkapex has joined #asahi-dev
yurinnick has quit [Quit: Connection closed for inactivity]
<chadmed>
sven: ive had a blog post about audio waiting around to be finished for like 3 months :P
<chadmed>
it just keeps growing, so i end up scrapping half of it as being indulgent and irrelevant