17 comments

[ 3.3 ms ] story [ 30.2 ms ] thread
Are these ARM CPUs much more power efficient than Intels chip? I have an ipad pro 11 and frankly in term of battery time, it feels about the same as a regular modern laptop.
I haven't seen data for the A14 big cores, but in Anandtech's Tiger Lake deep dive, they did make a comparison between the power usage of Apple's A13 cores and Tiger Lake.

>Comparing it against Apple’s A13, things aren’t looking so rosy as the Intel CPU barely outmatches it even though it uses several times more power, which doesn’t bode well for Intel once Apple releases its “Apple Silicon” Macbooks.

https://www.anandtech.com/show/16084/intel-tiger-lake-review...

Aren't they comparing to something a few years old now?

Looks like even the midrange AMD Zen3 stuff beats this...

The A14X is a part for tablets that have very little cooling capacity and run on battery most of the time. It's not really comparable to the 5600X that has higher power consumption than the entire iPad Pro the chip will end up in.
Microsoft's newly released Surface Laptop Go: (The most expensive SKU with 4 Core, 8 thread Ice Lake) - 1192 Single Core & 3305 Multi-Core.

https://arstechnica.com/gadgets/2020/11/microsoft-surface-la...

A14X: (4 big cores and 4 little cores) - 1634 Single Core & 7220 Multi-Core.

If accurate, it looks like a theoretical Macbook Air with this next gen iPad Pro chip would be more than competitive with Microsoft's offering.

It's geekbench... hold your horses.
The Surface Laptop Go is cheaper than an A14-powered iPad Air with a keyboard attachment. I don’t think any theoretical A14X-powered MacBook Air will be in the same price range as the Surface Laptop Go.
The iPhone SE, Apple Watch SE, and entry level iPad all offer a relatively inexpensive entry level option to their respective product lines.

Is there any reason to believe Apple won't apply the same pricing strategy to an Apple Silicon Macbook Air?

They might and it would be great if they did. But don't you think that such a laptop would receive the A14 rather than the A14X?

The reviews for the Surface Go Laptop are hilariously bad, as they are reviewing if it were intended to be a gaming or power user device. That's insane. It's a cheap, small, touchscreen laptop for doing schoolwork and writing Word documents - and probably the best one you can buy in that price range. I got one for my kids; once you uninstall the god-awful Skype and Mail apps, it runs silently all day long. The 3:2 screen makes writing Word documents much nicer than any similarly-sized widescreen laptop. Etc.

Anyway, I don't think that the MacBook Air would be positioned with this is a competitor. Apple tends to avoid this price range.

The initial Bloomberg leak that predated any announcement from Apple said that the entry level laptop chip would feature 8 big cores and 4 little cores and not 4+4 arrangement in this iPad Pro version of their chip.

Moving down to the iPad Pro version of the chip, given it's more than adequate performance, would already leave more pricing wiggle room.

Great news for all the heavy GeekBench users! Your workloads will run really fast on this CPU.

Reminder, Ipad Pro is apparently faster in multi-threading than 48 thread ZEN EPYC chip.

https://browser.geekbench.com/ios_devices/58

https://browser.geekbench.com/processors/amd-epyc-7401p

Is there a catch here? Or was PA Semiconductor some kind of superstar design company that's just way better than the big guys?
Geekbench isn't a perfect benchmark, and I dislike certain programmer keep banging on about it. But this wouldn't even be a fair comparison. The EPYC here is on GF 14nm, 2-3 Generation behind TSMC 5nm, ( depending on how you count them ), First Gen Zen Core IPC was mediocre in its time, and complicated NUMA design that makes Multithreading not performing at its best.

On a Zen 3 Ryzen 12 Core [1], again on a node that is one generation behind. But with much higher TDP allowance.

[1] https://browser.geekbench.com/v5/cpu/4597739

With the thermal and power throttling on that i9 it ain't much of an i9.
I know there are a lot of gimmicks where you can get outstanding performance if your workload fits within a cache.

Feels like maybe we're getting the same sort of trickery with thermals. I've seen a lot of over-the-top benchmarks lately from low-power devices, but if you tell them to run the task in a loop for half an hour, they saturate their heatsinks, and drop the clock to the sustained "non-turbo" clock or even throttle below it, so the score on the 10th run is a fraction of what it was on the 1st.

On the one hand, this is obviously gaming the knowledge that many benchmarks are short and bursty, but it makes me wonder if there's merit in trying to design a thermal solution that would allow permanent boost clock for thin-and-light-is-irrelevant tasks. If you had a iOS based kiosk or whatever, I could imagine installing a "shucked" iPad that was mounted to a much bigger cooler.