#
✕ nick:marcan
#asahi · 2021-09-23 16:18
marcan(chainload before running the script; once SMP starts chainloading breaks)
#asahi · 2021-09-23 16:18
marcanalso if you chainload the latest m1n1, the benchmark should reach 3.2
#asahi · 2021-09-23 16:17
marcanbut I want to see what bits it actually flips on older versions
#asahi · 2021-09-23 16:17
marcanmacOS 12.x does not touch it and even rewriting the same value to the register causes brokenness for me, so maybe it's a one-shot deal
#asahi · 2021-09-23 16:17
marcanmaz: can you pull and run cpu_pstates again? I want to know what it says about DVMR
#asahi · 2021-09-23 16:16
marcan:)
#asahi · 2021-09-23 16:01
marcanbut somehow writing to them breaks it :D
#asahi · 2021-09-23 16:01
marcanalso the strange thing is I'm not setting any bits in those registers... they're all already set for me
#asahi · 2021-09-23 16:00
marcanwhat was it before?
#asahi · 2021-09-23 15:59
marcanso apparently this works on <15.0 and breaks on 15.0 :D
#asahi · 2021-09-23 15:59
marcanbut only after this commit just now!
#asahi · 2021-09-23 15:59
marcanbizarre
#asahi · 2021-09-23 15:59
marcanfor some reason *that* CPU fails to start
#asahi · 2021-09-23 15:59
marcanTTY> Starting CPU 4 (1:0)... Failed!
#asahi · 2021-09-23 15:59
marcanmeanwhile this is now crashing for me :D
#asahi · 2021-09-23 15:59
marcanyeah, I fixed this recently :)
#asahi · 2021-09-23 15:58
marcanprior to the smp mode rework
#asahi · 2021-09-23 15:58
marcanyeah, so old m1n1 I'm guessing?
#asahi · 2021-09-23 15:58
marcan3G in m1n1 could mean an old m1n1
#asahi · 2021-09-23 15:57
marcan(until there's a cpuidle driver)
#asahi · 2021-09-23 15:57
marcanwhich will never happen in linux
#asahi · 2021-09-23 15:57
marcanj_ey: only with a single core active
#asahi · 2021-09-23 15:53
marcanforce pushed ;)
#asahi · 2021-09-23 15:52
marcanserves me right for not testing
#asahi · 2021-09-23 15:52
marcanahahaha
#asahi · 2021-09-23 15:51
marcanjust that script, I uncommented some stuff
#asahi · 2021-09-23 15:50
marcanif you didn't pull the update just now
#asahi · 2021-09-23 15:50
marcanyeah
#asahi · 2021-09-23 15:45
marcanalready gathered all the info (measured latencies too)
#asahi · 2021-09-23 15:45
marcaneither way, I'll see about poking at the cpufreq driver tomorrow, should be pretty easy
#asahi · 2021-09-23 15:44
marcan(mine doesn't, other people's do apparently)
#asahi · 2021-09-23 15:43
marcanI'm actually not sure what machines/versions need that, but whatever
#asahi · 2021-09-23 15:43
marcan(well, it's in beta now)
#asahi · 2021-09-23 15:43
marcanthen definitely not, it wasn't out yet :)
#asahi · 2021-09-23 15:42
marcanjust pushed a fix for that
#asahi · 2021-09-23 15:42
marcanactually if you're not on 12.x it might not quite be max
#asahi · 2021-09-23 15:41
marcanthat'll put all the big cores at max
#asahi · 2021-09-23 15:40
marcanif you're booting via usb/proxyclient, run experiments/cpu_pstates.py before
#asahi · 2021-09-23 15:40
marcanare you running with upclocked CPUs?
#asahi · 2021-09-23 15:40
marcan:D
#asahi · 2021-09-23 15:35
marcanfaster IPIs/IRQs
#asahi · 2021-09-23 15:35
marcanOTOH running *Linux* in a VM on Linux should already be faster than running it on macOS, since we use the M1's vGIC and they don't ;)
#asahi · 2021-09-23 15:34
marcan(on arm64 systems)
#asahi · 2021-09-23 15:33
marcanso I don't really count on graphical macOS ever being practical in a VM on Linux
#asahi · 2021-09-23 15:33
marcanvirtualizing macos is a bit of an issue, you'd have to run the vmkernel version of macos at least and GPU support requires writing a Metal implementation
#asahi · 2021-09-23 15:32
marcanyes, fex is what I intend to provide in our images
#asahi · 2021-09-23 15:21
marcanand Linux on M1 is going to be much higher impact than Linux on PS3 ever was
#asahi · 2021-09-23 15:20
marcanit is literally illegal to do this
#asahi · 2021-09-23 15:20
marcanand should've also been sued under other laws (e.g. european warranty laws)
#asahi · 2021-09-23 15:20
marcanremember, Sony lost a class-action due to this

plus ›