Hacker News Reader

Comment by azornathogron | Hacker News Reader

How much of the Mac's impressive battery life is due purely to CPU efficiency, and how much is due to great vertical integration and the OS being tuned for power efficiency?

It's a genuine question; I'm sure both factors make a difference but I don't know their relative importance.

View on HN

I just searched for the asahi linux (Linux for M Series Macs) battery life, and found this blog post [0].

> During active development with virtual machines running, a few calls, and an external keyboard and mouse attached, my laptop running Asahi Linux lasts about 5 hours before the battery drops to 10%. Under the same usage, macOS lasts a little more than 6.5 hours. Asahi Linux reports my battery health at 94%.

[0] https://blog.thecurlybraces.com/2024/10/running-fedora-asahi...

The overwhelming majority is due to the power management software, yes. Other ARM laptops do not get anywhere close to the same battery life. The MNT Reform with 8x 18650s (24000mAh, 3x what you get an MBP) gets about 5h of battery life with light usage.

The MNT Reform does not use Li-Ion batteries, resulting in much poorer energy density. It's going to depend on the cells being used but this is what I could see from the Reform Next detail page: 8× LiFePO4 cells (16000 mAh total). Assuming 2200mah cells, I think this nets you around 56 Wh in a 4S2P configuration, Li-Ion cells would be closer to 65 Wh.

According to Apple's website it seems a 14 inch macbook pro has a 70 Wh battery.

Other ARM laptops also have much worse SoCs, so it's not really apples-to-apples.

As someone else posted elsewhere in this thread, Asahi nets about 6.5 hours on a Macbook. Apples to apples the story is the same. It's really not that shocking. The effect of power management (or rather, when it's lacking) is very apparent when you do bare metal programming. The first thing you do when you finally get something booting is kill a battery in less than an hour with a while(true).

I think you may have misread that post. It said Linux was 5 hours, MacOS was 6.5 hours. There's a gap (Linux gets 77% the life of MacOS), but it's apparently not that significant.

Tight integration matters.

Look at the difference in energy usage between safari and chrome on M4s.

How much instruction perf analysis do they do to save 1% (compounded) on the most common instructions

it's less that and more all the peripheral things. USB, wifi, bluetooth, ram, random capacitors in the VRM etc.

Reply on news.ycombinator.com