Activity Monitor in macOS gets Apple Silicon’s power consumption wrong


AppleInsider is supported by its audience and is eligible to earn an Amazon Associate and Affiliate Partner commission on qualifying purchases. These affiliate partnerships do not influence our editorial content.

Activity Monitor in macOS may not be as accurate with the data it provides to Apple Silicon users, with a report claiming that the tool cannot correctly differentiate between performance and power cores. efficiency.

Activity Monitor gives users and developers a way to find out which apps are using the most resources and the most power when performing tasks. While testing elements of the tool’s functionality on a Mac running Apple Silicon, it appears that a small error in kernel recognition may have skewed some of the results.

Using the CPU and Energy numbers from the Activity Monitor, a user can potentially see that an application’s code running only on efficiency cores apparently uses more energy than performance cores. when performing a task. As efficiency cores are supposed to be slower but with lower power consumption than performance cores, the result is quite contradictory.

Tested by The electric lighting company on a Mac Studio with an M1 Max involved running threads on all 8 performance cores and 2 efficiency cores. When 8 floating-point math threads with 1 billion loops per thread are throttled to each core type, the 8 performance threads completed the task in 6.6 seconds and the 2 efficiency cores took 40.4 seconds .

However, checking the Energy tab of the Activity Monitor indicated that the performance cores maintained an energy value of 800, with a total of 5280 units used at 660 per thread. Meanwhile, the efficiency cores maintained a power value of 194, with a total of 7,838 units consumed, or 980 per thread.

Taken at face value, this would imply that running those specific threads on efficiency cores turns out to be less efficient than performance cores. The problem is caused by Activity Monitor, according to the report, as it cannot differentiate between identical cores with a fixed frequency and two different types of cores with varying frequencies.

It was also found to have issues with the way it reported the load between cores. A determined test running twice the amount of code on two efficiency cores was reported in Activity Monitor as using the same amount of power as half the amount of code.

The results are also believed to be inconsistent between different M1 chips due to the way macOS controls the frequency of the efficiency cores, so an M1 will handle reports differently than an M1 Pro, for example.

“Until Apple updates the numbers returned by Activity Monitor for M1 chips, confusion by core type and frequency renders it not only useless, but actually misleading for comparing CPU percentage or energy .” adds the report. Instead, tools like Powermetrics should be considered by developers, as they provide information on cluster frequencies and power consumption, as well as active residency.

It’s unclear whether Apple intends to update Activity Monitor. It hasn’t seen any notable improvements since Apple Silicon has been around.


Source link

Steven L. Nielsen