#asahi · 2023-08-16 15:43
_jannau__you could do a minimal installation and switch the root partition fs type from the initramfs with rd.break#asahi · 2023-08-16 15:38
_jannau__however the fedora based images use btrfs for the root partition, due to be released soon#asahi · 2023-08-16 15:38
_jannau__pounce: not easily you would have to recreate the disk image used by the installer. The installer runs on macos and has no tools or fs drivers for linux file systems#asahi · 2023-08-11 07:46
_jannau__yes#asahi · 2023-08-11 07:21
_jannau__alexander: yes, all models use the same chipset, some Marvell/Aquantia 10gbit one, maybe something with 107#asahi · 2023-08-09 12:21
_jannau__half working. the nomodeset disables display and gpu driver. the latter not fully but for dispaly purposes#asahi · 2023-08-09 12:09
_jannau__not sure if that is helpful#asahi · 2023-08-09 12:05
_jannau__stintel: has your config apple_dart as built-in or as a module?#asahi · 2023-08-09 12:04
_jannau__those duplicate dart sysfs where iirc also visible in rob's and my starnge dart errors seen many months ago#asahi · 2023-08-09 12:02
_jannau__the alx isntaller lists only asahi images with 12.3/4#asahi · 2023-08-08 16:52
_jannau__it has to be something in the initramfs/kernel. m1n1 can't write to the ESP and macOS ignores it as well especially random files (to its knowledge)#asahi · 2023-08-08 16:49
_jannau__does it stick if you unmount and mount the esp partition#asahi · 2023-08-08 16:47
_jannau__stintel: you mount ESP in macos, replace m1n1/boot.bin reboot into your uefi-only install which complains about a missing devicetree for your device? after rebooting to macos you determine m1n1/boot.bin is not yours. how do you determine that? by size? by sha156?#asahi · 2023-08-04 13:57
_jannau__that's too late. can you check whether the module is in the initrd? I'd guess it's not but it probably should be#asahi · 2023-08-04 13:51
_jannau__do you have macsmc-rtc log messages? is this with packaged kernels from asahi or fedora?#asahi · 2023-08-03 14:30
_jannau__unofficially: backup the ESP (and your data), delete stub and efi partition, make an uefi-only install with 13.5, restore the ESP content and deal with the changed part uuid of the ESP#asahi · 2023-08-03 14:27
_jannau__tpw_rules: no official but there is no need for it (for now). touchbar firmware on early installations would be an exception#asahi · 2023-07-31 13:16
_jannau__it's possible (but unlikely) that 30 core's can't use use the full 400 GB/s bandwidth, all CPUs cores on the m1 max have less than 400 GB/s mem bandwidth#asahi · 2023-07-31 13:12
_jannau__I suppose only offering 96GB with the full GPU is a way to keep the number of configs down#asahi · 2023-07-31 13:10
_jannau__memory bandwidth just depends on the chip, not on the amount of ram#asahi · 2023-07-31 13:09
_jannau__if you upgrade a mac mini m2 pro to 12 CPU cores + 32GB ram, you can by the mac studio m2 max for the same price#asahi · 2023-07-31 13:08
_jannau__for mostly CPU-only workloads there should be no difference, if you start using every single unit on the SoC at the same time there might but I'm not sure how much better the mini's cooling system it#asahi · 2023-07-26 12:38
_jannau__tamar: probably a kernel vs. dtb mismatch. did you run update-m1n1 or otherwise ensured that the 2nd stage m1n1 has dtbs that match your kernel#asahi · 2023-07-26 08:13
_jannau__Retr0id: ignore the dkms the kernel has CONFIG_WIREGUARD=m#asahi · 2023-07-24 07:13
_jannau__the errors sounds more like qemu was built without kvm support. Where is that binary coming from?#asahi · 2023-07-12 08:46
_jannau__the asahi-installer will only work under macos. caffeinate is the least of your problems#asahi · 2023-07-06 11:50
_jannau__https://github.com/AsahiLinux/docs/wiki/Introduction-to-Apple-Silicon#firmware-overview#asahi · 2023-07-06 11:49
_jannau__I think the avd firmware is embedded in the macos driver#asahi · 2023-06-29 13:02
_jannau__isn't it pytorch nowadays?#asahi · 2023-06-29 10:54
_jannau__I don't think ANE has anything to do with metal, you want to look at https://github.com/eiln/anecc and https://github.com/eiln/ane#asahi · 2023-06-27 11:54
_jannau__a fixed u-boot should be available before the devices are fully supported. patches are available. a re-install will be required for display/gpu support since we will only support firmware from different macos version#asahi · 2023-06-27 11:04
_jannau__the m2 pro/max devices are not yet supported. keyboard support in u-boot breaks the device so it's unuseable in linux#asahi · 2023-06-27 10:28
_jannau__mooad_dib: no, the installing is always necessary. even when full usb storage support is implemented, there will be still an install step#asahi · 2023-06-27 09:56
_jannau__gcc man page says that sb, ssbs, predres do not affect code generation just instruction availability for the assembler#asahi · 2023-06-27 09:35
_jannau__at least for some of them like bti I can understand why they are not turned on by default for armv8.5-a based CPUs, for others I can only assume they are trying to save bits. the arch extensions is a 64-bit bitmask and almost full#asahi · 2023-06-27 09:32
_jannau__psykose: llvm differentiates between arch extensions and cpu features. the mandatory new armv8.5-a features like bti/m2/frintts have just feature flags#asahi · 2023-06-27 09:00
_jannau__the annotation of M1 as Armv8.5 is wrong, the problem is that there's no difference between ARMV8_4A and ARMV8_5A as far as llvm is concerned: https://github.com/llvm/llvm-project/blob/9a4097e9cd891e1445ba07ac6a333321fe117fce/llvm/include/llvm/TargetParser/AArch64TargetParser.h#L323#asahi · 2023-06-27 08:53
_jannau__whyirc: the Arm architecture reference manual. see chapter A2 "Armv8-A Architecture Extensions"#asahi · 2023-06-27 08:52
_jannau__m2 supports BTI, nested virtualisation and some other additional features#asahi · 2023-06-27 08:51
_jannau__m1 and m2 differ. Mx and Mx Pro/Max do not differ, their cores share the same micro arch#asahi · 2023-06-27 08:50
_jannau__whyirc: MTE is optional in armv8.5#asahi · 2023-06-19 14:16
_jannau__sid-linux: more likely than not is that following the instructions broke it. We had that in the past. Not sure if the underlaying cause was resolved. iirc there's a debian like update-grub which avoid some pitfalls#asahi · 2023-06-19 14:13
_jannau__Foxboron: thanks, seems to be in place now#asahi · 2023-06-14 11:42
_jannau__whyirc: afaik upower tries convert Ah to Wh and fails#asahi · 2023-06-14 08:38
_jannau__the laptop displays are all limited to 60Hz max#asahi · 2023-06-14 08:37
_jannau__it all depends on the display#asahi · 2023-06-14 08:24
_jannau__x11 applications under x11/modeset can update the framebuffer more than 60 times per second, see broken animation timing on plasma startup for exampel#asahi · 2023-06-14 08:22
_jannau__if xrandr reports 60Hz the display is refreshed with 60Hz. that has has however little to do with that happens in X11. There is nothing to limit screen updates to the display refresh rate#asahi · 2023-06-14 08:00
_jannau__bcrumb: I don't understand the question. it is whatever the display supports and is currently configured#asahi · 2023-06-14 07:51
_jannau__changing the mode is compositor dependend but the compositor config should report the refresh rate as well