#asahi-dev · 2026-06-04 09:31
jannaum3 pro looks better but I doubt 8.784 vs 10.570 coremark iterations/s per MHz are expected #asahi-dev · 2026-06-04 09:09
jannaudhrystone#asahi-dev · 2026-06-04 09:09
jannauI'm using coremark instead whrystome though since 40 year old C code is annoying#asahi-dev · 2026-06-04 09:08
jannauDocumentation/devicetree/bindings/cpu/cpu-capacity.txt#asahi-dev · 2026-06-04 09:07
jannauI'm currently doing it on m3 and and the perf cores are slower than the efficiency cores (both at max frequency)#asahi-dev · 2026-06-04 09:06
jannauyuka, chaos_princess: did you already benchmark cpu cores for capacity-dmips-mhz on m3 max?#asahi-dev · 2026-06-04 07:49
jannauyuka: yes, just half read what you said then#asahi-dev · 2026-06-04 06:20
jannauthose == the apple counter sysregs#asahi-dev · 2026-06-04 06:20
jannauyou could try if we can trap those. there are unknown bits in HACR_EL2 and AHCR_EL2 might be another hyp config reg#asahi-dev · 2026-06-04 06:19
jannauwe probably need to fill a radar about CYC_OVRD as well#asahi-dev · 2026-06-03 22:03
jannauand the recently "fixed" chainload.py#asahi-dev · 2026-06-03 21:44
jannauit wasn't intended to switch the cpu core and keep using m1n1 but only for kboot#asahi-dev · 2026-06-03 21:43
jannauit's possible that there is still code checking cpuX.state == "running" to determine the boot cpu#asahi-dev · 2026-06-03 21:03
jannaubluetooth audio issues confirmed on 7.0.11, bt trackball works though#asahi-dev · 2026-06-03 21:02
jannaucy8aer: I switched brcmext patch from v2 to v3. can you try to revert 30fcc498ff7c and apply f5cd0acdc21d#asahi-dev · 2026-06-03 20:20
jannaunot too late to change that#asahi-dev · 2026-06-03 18:39
jannauon the max you could test a core on the second perf cluster. I suspect switching to an efficiency core at the beginning will be sufficient. that probably explains why we did not see it on m1* and m2*#asahi-dev · 2026-06-03 18:22
jannaulooks good now#asahi-dev · 2026-06-03 18:21
jannaudoes the benchmark gives you the expected frequency for higher p-states starting around 3.7 GHz#asahi-dev · 2026-06-03 18:18
jannauyuka: something went wrong in the cpufreq PR: "yuyuyureka wants to merge 3290 commits into AsahiLinux:bits/001-devicetree-m3 from yuyuyureka:t6031-cpufreq"#asahi-dev · 2026-06-03 07:01
jannauit's not required to create frame pointers for tail calls. as long as the link register is not disturbed you can still return#asahi-dev · 2026-06-02 20:53
jannauprobably close enough to to j514. the missing cutout for the usb-c port on the right side hopefully make not much of a difference#asahi-dev · 2026-06-02 20:36
jannauI've tagged and pushed asahi-7.0.11-1 (bare minimum to get all 14.x based m3 devices to be self-updatable). depends on m1n1 t603x pcie support#asahi-dev · 2026-06-02 19:06
jannaushould be /usr/share/alsa/ucm2/conf.d/macaudio#asahi-dev · 2026-06-02 19:06
jannau`equery f alsa-ucm-conf-asahi`#asahi-dev · 2026-06-02 18:57
jannauwhen does this happen? on use? Message sounds familar. Did you adjust the ucm to cover M3 devices as well? "or UCM profile" might be true#asahi-dev · 2026-06-02 17:00
jannaudidn't dump_pmgr output it on m1/m2?#asahi-dev · 2026-06-02 16:47
jannau20 probably needs the other cpu cores in idle but 17 and 18 look wrong#asahi-dev · 2026-06-02 16:46
jannau... 15:3624MHz 16:3708MHz 17:3705MHz 18:3704MHz 19:3960MHz 20:3959MHz#asahi-dev · 2026-06-02 16:46
jannauthe 20 p-states on everest behave a little strange in the upper range:#asahi-dev · 2026-06-02 16:45
jannauare you missing e-core pstates? t6030 has 8 up to 2748 MHz#asahi-dev · 2026-06-02 16:30
jannauyeah: "current CPU frequency: 10.44 GHz"!#asahi-dev · 2026-06-02 16:22
jannauI want "opp-mhz", annoyinmg to keep track of 6 trailing zeros#asahi-dev · 2026-06-02 16:18
jannaudoes the t6031 table matches powermetrics? both t8122 and t6030 are different then. both have 700 - 800 MHz p-states for everest and sawtooth (I guess the non working op_01 p-state)#asahi-dev · 2026-06-02 16:01
jannau20 distinct p-states according to powermetrics#asahi-dev · 2026-06-02 16:00
jannauno, even base M3 has 4056 MHz as highest p-state#asahi-dev · 2026-06-02 15:56
jannauI don't think we have the high perf ones yet. highest is above 4 GHz#asahi-dev · 2026-06-02 15:54
jannau14-inch m3 max should have it as well#asahi-dev · 2026-06-02 15:53
jannauhttps://support.apple.com/en-us/101613#asahi-dev · 2026-06-02 15:53
jannauand 16-inch m1 max#asahi-dev · 2026-06-02 15:52
jannau14-inch but didn't apple extended performance mode to all models? it was only the 16-inch m2 max iirc#asahi-dev · 2026-06-02 15:42
jannaucpufreq works on t6030 as well. so much smoother. nice#asahi-dev · 2026-06-02 11:19
jannaunot sure if there is a good way to model this. what possible could work is to runtime suspend avd before the reset. That should runtime suspend the dart as well which saves its state and restores it when it goes out of suspend#asahi-dev · 2026-06-02 11:17
jannaumy best guess is that the dart loses stated due to the reset. if you run under the HV you could check#asahi-dev · 2026-06-02 11:10
jannauwhich power-domains do your avd and dart_avd devices have?#asahi-dev · 2026-06-02 11:05
jannauis that a videobuf address?#asahi-dev · 2026-06-02 11:04
jannauwhat does reset mean? reset while the module remains loaded and the device bound?#asahi-dev · 2026-06-01 19:53
jannauhttps://asahilinux.org/docs/sw/m1n1-hypervisor/ has instructions#asahi-dev · 2026-05-31 09:32
jannaufixed#asahi-dev · 2026-05-31 09:28
jannautypo in t6034-j516m.dts