Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

The most interesting thing to me is that switching to a new SOC process node gives you the option of bumping up performance or reducing power draw, and it sounds like Apple has once again chosen to reduce power draw.

It seems like other vendors have been going the "push performance no matter what it does to power draw and heat" route for a long time now.




On the CPU game, apple is so far ahead that it might not even make sense to talk about other vendors. Apple could deliberately downclock their CPUs and still beat the shit out of any flagship android. Hell, their 3 years old lineup actually beats the current year’s top.


It mostly comes down to manufacturing processes. When you compare 5nm Apple hardware to 5nm hardware from other competitors, the power/performance disparity closes significantly. Since Apple is the biggest though, they can out-spend anyone else who tries to buy out the latest node right now. This has been the case since 2020, and seems to still hold true.


Just plain not true. Process is relevant but is not the biggest factor. Design is.


Design makes a big difference, but Apple is not the only company designing high-performance ARM cores on TSMC's finest. Some might even say they aren't the best designers doing it, either.


Some might


I imagine the reasoning is that most people do not use the performance, but would value less need to charge the battery.


Your conclusion is wrong. There is no difference in battery life if you compare to 14 Pro on their site.


More than one component in the device draws power. Did you see the part about the display getting twice as bright, for instance?


15 Pro has the same brightness specs as 14 Pro.


Their website is not a reliable source for performance numbers. We'll have to wait for third party benchmarks


I disagree. Apple is the only site that gives realistic numbers rather than theoretical maxes.

Independent benchmarks are good, but Apple keeps everything very focused on what an average person buying it would experience, even benchmarks.


Why run faster if you are already comfortably number one? It makes sense to focus on something that everyone wants improvement on: battery life.


To be honest, in general it does improve battery life if you are faster. Battery life is a race to CPU sleep mode, and if you finish the job faster, the more energy you can save.


Only for things that are "completable", not for things that run in the background persistently


Not much runs persistently, more like periodically. And the amount of time it has to be running still matter — processors when turner on leak a bunch of energy no matter what.


I’m skeptical about this. Why would a faster processor in it of itself be more energy efficient at executing instructions?


A faster processor reduces execution time allowing the CPU to return to sleep status faster. Total energy used is power x time. While increasing operating frequency increases power quadratically due to P = CV^2, it is usually a net energy savings to let the frequency boost as high as a mass produced chip can go to minimize time spent in the highest power state. This assumes a finite bursty workload.


>and it sounds like Apple has once again chosen to reduce power draw

We will have to see. They have a 10% performance increase and I am betting there will be 5% coming from Clock Speed.

And it is not like Apple could push power draw any further as they have maxed out single core power draw. Unlike other competitor ( ARM / Snapdragon ) which has a much lower single core power draw figures and hence they try to push as much as possible.


> We will have to see. They have a 10% performance increase and I am betting there will be 5% coming from Clock Speed.

The big cores got faster, but the little ones did not.

> the performance cores are 10 percent faster... the efficiency cores are more efficient rather than being faster

https://arstechnica.com/gadgets/2023/09/apples-new-iphone-15...

Which sounds to me like a new version of the big cores, but not of the little cores on a more power efficient process node.

However, as always, time will tell.


Well it turns out it is even worst [1], little to no IPC increase and purely clock speed increase. The CPU is basically a die shrink of A16.

Looks like any IPC will likely be in A18.

[1] https://www.macrumors.com/2023/09/14/iphone-15-pro-geekbench...




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: