m0xEE on Nostr: So, disabling inactive (no devices connected) Bluetooth with rfkill and explicitly ...

So, disabling inactive (no devices connected) Bluetooth with rfkill and explicitly bringing down also already inactive (no cable plugged in) wired network interface with "ip link set down dev (interface_name_dreamt_up_by_poettering)" enabled me to reset BD_PROCHOT (unrelated to prosciutto, it means "bi-directional processor ♨️" in Intel-speak) in MSR, which Mac EFI sets if when you boot the thing without battery — to prevent it from crossing the threshold of power input that adapter is able to provide and avoid unexpected shutdowns.

And what can I say, now I can run this i7 clocked at 2,4 GHz — its designated frequency! This is what I observe over SSH session — yes, I've made a "hardware performance monitor" out of feces and sticks (continuous reads from /sys with cat 🫢), all of its real cores and their HitlerThreading-spawned evil twins run at 2400MHz, and its both fans went slightly beyond their limit of 5500 RPM. I'm not sure how accurate powertop readings are, but CPU power consumption is currently at ~28W, I've seen it go as high as 32W while building GCC — still not even close to its theoretical limit, which according to Intel datasheet is 45W. Perhaps due to the fact that I have TurboBustier👙™ disabled — I'm not sure sacrificing stability for the option to have these single-core "bursts" is a good idea. Another contributing factor could be the fact that its display is off and I believe that power draw of its integrated Intel GPU is accounted for in CPU TDP.

First lesson to learn here is that life without battery and infamous BD_PROCHOT hard-locking frequencies at 800MHz is possible! Playing back 4k videos for which it doesn't have hardware decoding capability can still cause it to power off. Contributing factors could be integrated GPU activity, audio output, slightly different CPU workload and even something as obvious as display backlight 😏
But normal workloads are okay: I've already built Firefox for three architectures and currently in the process of cross-building LLVM for Big Endian 64-bit PowerPC — seems very stable!

And the second one, again I'm not sure how accurate PowerTop readings are, and maybe integrated Intel ethernet adapters are better in this regard, this MacBook has a Broadcom one handled by tg3 kernel module.
But wired interface without any cable even plugged in drawing almost as much as panel backlight at medium brightness, substantially more that wireless adapter connected to a network and actually transmitting data — wow, that's some exceptionally bad power management! 😲
Instead of improving it — making it even remotely as good as it is for wireless networking, they just got rid of integrated wired network adapters in most laptops! 😩