Apple M2 heating problems - in my view proof of Arm bad architecture

[smeyer0028]

Here is story from Arstechnica on over heating problems in new Apple Macbooks.

The new MacBook Air runs so hot that it affects performance. It isn’t the first time [Updated]

Apple can (and probably should) provide more cooling for the M2 MacBook Air.

arstechnica.com

I am not sure of the significance of the story. Is Apple using the new speed enhanced TMSC process
for its new M2 chip? I have argued this before and I think it was understood by John von Neumann
in his original computer design that more complex addressing increase computing power. Arm chips
at same fab parameters will therefore have performance problems. Maybe Apple can no longer compete
on just lower power. Does anyone know more about this?

 

[Xebec]

I don't follow. The M2 "running hot" is an engineering choice (both in chassis design - staying fanless) and in performance design of the chips on the PCB. It has nothing to do with the instruction set.

The metric to consider is performance per watt though even that only lightly correlates to ISA as even x86 doesn't use too much power to translate instructions to an internal risc set. (David Kantar has posted some good info on this).

Apple simply chose to push the clocks up more this time and allow the CPU to opportunistically use them when chassis cooling was sufficient. They then took a few cost shortcuts on the thermal materials used inside the Air to manage those thermals.

 

[jms_embedded]

They then took a few cost shortcuts on the thermal materials used inside the Air to manage those thermals.

Sophisticated real-time thermal management and control sounds right up Apple's alley. I interviewed there about 10 years ago and that was one of the areas that caught my interest. (But not as much as motor control does. )