Apple's A15 Chip Faster Than Company's Own Claims in Independent Tests

[Frantisekj]

I think this test answers why is A15 multicore result so good even power core is not that much better and also why are android phones sometimes speedier in app launch and other tests but are far from iPhone power efficiency.

A15 efficiency cores are now much more powerful (-> higher multicore results) but still less powerful then SD888 there fore androids can be faster as I think efficient cores can be used for launching apps (just my guess)
But A1X were ever much more power efficieant so it is reason why iPhone with 4000-4500 batteries would run rounds around androids.

 

[leman]

Citation for 2 and 3?

With pleasure!

Regarding 2:

https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=8257U

Even in low power mode, the iPhone 13 Pro A15 Bionic chip still embarrasses the competition

It is well-known that Apple has a significant lead over the competition in terms of both the performance and efficiency of its Arm-based designs. However, even in low power mode, Geekbench 5 shows that the A15 powering the Apple iPhone 13 Pro is still powerful enough to embarrass the competition.

www.notebookcheck.net

Regarding 3:

GFXBench - Unified cross-platform 3D graphics benchmark database

The first unified cross-platform 3D graphics benchmark database for comparing Android, iOS, Windows 8, Windows Phone 8 and Windows RT capable devices based on graphics processing power.

gfxbench.com

 

[leman]

I think this test answers why is A15 multicore result so good even power core is not that much better and also why are android phones sometimes speedier in app launch and other tests but are far from iPhone power efficiency.

A15 efficiency cores are now much more powerful (-> higher multicore results) but still less powerful then SD888 there fore androids can be faster as I think efficient cores can be used for launching apps (just my guess)
But A1X were ever much more power efficieant so it is reason why iPhone with 4000-4500 batteries would run rounds around androids.

Launching apps has little to do with the CPU performance and a lot with the OS itself. If Apps launch faster on Android, it’s because the OS is designed for faster app launch or maybe the app is not optimized for quick launch or both. Anyway, where did you get that from? Hot app launch on my 11 non pro is instant, how can one meaningfully do faster than that?

 

[dennisbrekke]

The thing that blows me away with the 13 Pro Max is that I can take a 4k video and through the Photos app, fine-tune the color, details, etc. and it takes ZERO time to post-process it when I click "done." On my 12 Pro Max, it took forevvvvvver to do any post-processing on videos, even in lower res. Maybe this is also an improvement in software, but WOW, I'm impressed.

 

[steve09090]

Just some fun facts:

• The iPhone CPU now has more cache than the state-of-the art desktop CPUs that cost almost as much as the entire iPhone itself
• Just the four efficiency cores in the iPhone are about as fast as the 2019 13" MacBook Pro
• The iPhone GPU is faster than the Iris Plus in the $1799 four-port 13" MBP

Yep, Apple is clearly unable to innovate and is entirely doomed...

Another fun fact:

The Anti-Apple Trolls usually on this site must have missed this thread. They’re a hurtin'

 

[ikjadoon]

Just some fun facts:

• The iPhone CPU now has more cache than the state-of-the art desktop CPUs that cost almost as much as the entire iPhone itself
• Just the four efficiency cores in the iPhone are about as fast as the 2019 13" MacBook Pro
• The iPhone GPU is faster than the Iris Plus in the $1799 four-port 13" MBP

Yep, Apple is clearly unable to innovate and is entirely doomed...

With pleasure!

Regarding 2:

https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=8257U

Even in low power mode, the iPhone 13 Pro A15 Bionic chip still embarrasses the competition

It is well-known that Apple has a significant lead over the competition in terms of both the performance and efficiency of its Arm-based designs. However, even in low power mode, Geekbench 5 shows that the A15 powering the Apple iPhone 13 Pro is still powerful enough to embarrass the competition.

www.notebookcheck.net

Regarding 3:

GFXBench - Unified cross-platform 3D graphics benchmark database

The first unified cross-platform 3D graphics benchmark database for comparing Android, iOS, Windows 8, Windows Phone 8 and Windows RT capable devices based on graphics processing power.

gfxbench.com

Just to note, I don't think that "low power mode only uses the small cores" has ever been verified.

AnandTech's own benchmarks are much more accurate. From their SPEC2017 Geomean database (none of which are a MacBook Pro 2019, but it's likely between 5.5 to 6 points):

• Intel i7-1065G7 (Surface Laptop 3):7.0 points
• AMD Ryzen 4800U (Yoga Slim 7): 6.1 points
• Intel i7-8550U (Matebook X Pro): 5.6 points
• A15 Blizzard Efficiency Core (sans the missing sub-test): 2.7 points

So the efficiency cores are likely still only 50% as fast as a MacBook Pro 2019. But, if you want a superlative for the list, you can add that the A15's Avalanche P cores score a higher SPECfp2017 (floating point) score than any other CPU ever tested at AnandTech, including the giant 5950X.

 

[aTomek]

X86 architecture is at the end of its road, Apple showed that ARM is the future. I have mixed feelings about this, as for 20+ years I was fan of PCs, but lack of innovation from Intel and great implementation of ARM ISA by Apple made me completely switch to Apple (M1). The only thing Apple is behind is gaming industry, but I hope it will change quickly with M1 adoption in game engines.

 

[leman]

Just to note, I don't think that "low power mode only uses the small cores" has ever been verified.

AnandTech's own benchmarks are much more accurate. From their SPEC2017 Geomean database (none of which are a MacBook Pro 2019, but it's likely between 5.5 to 6 points):

• Intel i7-1065G7 (Surface Laptop 3):7.0 points
• AMD Ryzen 4800U (Yoga Slim 7): 6.1 points
• Intel i7-8550U (Matebook X Pro): 5.6 points
• A15 Blizzard Efficiency Core (sans the missing sub-test): 2.7 points

So the efficiency cores are likely still only 50% as fast as a MacBook Pro 2019. But, if you want a superlative for the list, you can add that the A15's Avalanche P cores score a higher SPECfp2017 (floating point) score than any other CPU ever tested at AnandTech, including the giant 5950X.

You are right! Sorry, I had a total brain fart. The E-cores only offer 25-30% of the P-core performance, those low energy mode Geekbench scores are way too high to be E-cores alone. Another example why one shouldn’t turn off one’s brain when reading tech news articles…

Thanks for pointing this out!

 

[M3gatron]

Launching apps has little to do with the CPU performance and a lot with the OS itself. If Apps launch faster on Android, it’s because the OS is designed for faster app launch or maybe the app is not optimized for quick launch or both. Anyway, where did you get that from? Hot app launch on my 11 non pro is instant, how can one meaningfully do faster than that?

Opening/Launching apps has to do with single core CPU performance, I/O performance even Network performance as most apps connect to the internet and the OS and apps themselves. What you ignore is that 2 or 3x higher CPU performance doesn't mean 2 or 3x faster launch time or that apps are 3 times faster in general.

 

[Aaargh!]

The iPhone CPU now has more cache than the state-of-the art desktop CPUs that cost almost as much as the entire iPhone itself

To be fair, it’s an RISC CPU so it needs more cache to begin with.

 

[Whathappened]

This article is one of the reasons why I’ll continue to use Apple products for the foreseeable future.

Really? I couldn’t care less about performance gains. Even my spare iPhone XR is more than fast enough.
I stay with Apple because of the eco system and the awesome out of the box picture quality.

 

[Wizec]

To be fair, it’s an RISC CPU so it needs more cache to begin with.

Why?

”You get a bigger speed improvement by using CISC and caches than RISC and caches, because the same size cache has more effect on high density code that CISC provides.”

Why does Intel hide internal RISC core in their processors?

Starting with Pentium Pro (P6 microarchitecture), Intel redesigned it's microprocessors and used internal RISC core under the old CISC instructions. Since Pentium Pro all CISC instructions are divi...

stackoverflow.com

Intel and AMD chips are RISC.

Why does Intel hide internal RISC core in their processors?

Starting with Pentium Pro (P6 microarchitecture), Intel redesigned it's microprocessors and used internal RISC core under the old CISC instructions. Since Pentium Pro all CISC instructions are divi...

stackoverflow.com

 

[Shadows_lord]

Given the AnandTech chart, we should expect a multi-core Geekbench score of ~12000 for M1X.

 

[atomic.flip]

The problem hasn't been hardware or performance for a while now. It's about utilizing it. iOS or even iPadOS is simply not using all that performance. No pro apps. Locked down OS. At this point, all these advances in performance is almost a mockery, as if they want to say "Yes we have the best chip - but you can't really use it to its max potential".

This! Who gives a damn if your watch can travel at the speed of light when you can’t even use it as an alternative to your car because it’s only designed to operate as a time telling device.

 

[leman]

To be fair, it’s an RISC CPU so it needs more cache to begin with.

Where did you get this notion from?

 

[WWPD]

I hope Apple can improve the thermals in the next iPhone, what good is all that power when it has to be throttled because it gets too hot.

 

[DrV]

To be fair, it’s an RISC CPU so it needs more cache to begin with.

Why would that be?

There are a lot of comparisons between ARM64 and X86-64. Some of them state ARM code is a few percent larger, some of them claim ARM code to be significantly smaller. One way to compare code size is to have a look at some common Linux utilities with common source code for both architectures. The result in that comparison seems to be that ARM code is slightly more compact.

I would say the difference is negligible and depends on the use case. In general, old CISC code (e.g., 8086 8/16-bit) is compact compared to equivalent RISC machines, but the difference has diminished with current 64-bit processors. And even in the embedded world in certain tasks ARM-Cortex (32-bit RISC) may take less space than AVR (8-bit RISC) which has smaller footprint than 8051 (8-bit CISC). This is completely against the conventional wisdom, but it all depends on the architecture, compiler, and task.

The decades old ideas of RISC and CISC should probably be discarded. All fast CISC processors translate the instructions into RISC instructions before execution, as there is no viable way to run the parallel decoding and pipelining with "pure" CISC instructions. This has lead to the situation where the CISC code produced by modern compilers has become more akin to RISC code, as that gives faster execution.

The OS has also a significant impact on the cache use at different levels, as a lot of the code running may be shared library code. Apple controls the whole stack from each transistor in the CPU to the OS, which enables optimizing the whole chain.

 

[svish]

Wow! Apple chips are truly powerful

 

[now i see it]

The A15 speed is all fine and dandy but… it’s running buggy as all get go iOS 15.

 

[Sophisticatednut]

The Anti-Apple Trolls usually on this site must have missed this thread. They’re a hurtin'

Not at all, waiting for the apple M2 with exictment. Hopefully they will have dedicated GPU or support for nvidia and and cards

 

[leman]

Not at all, waiting for the apple M2 with exictment. Hopefully they will have dedicated GPU or support for nvidia and and cards

They won't. Why would you won't them to use slower GPUs?

 

[ArtOfWarfare]

vastly faster and give out a better gaming experience than competitive phones.

It's been forever since I heard about a AAA game coming to iPhone... or any game, really.

I hear that Apple keeps making money off games on the iPhone, but I never hear about games actually being any good on the platform. Obviously good graphics and good games aren't the same thing, but I don't hear about either on the iPhone.

 

[dugbug]

I hope Apple can improve the thermals in the next iPhone, what good is all that power when it has to be throttled because it gets too hot.

sustained performance for the A15 is still greater than the max/peak on the other chipsets. Its not really that bad. Anandtech was just pointing out an area where improvements can be made.

 

[840quadra]

Impressed with early tests on my iPad mini that features the under clocked version of this processor. Just in my normal compute tasks it is faster than my iPad Pro from a couple years ago. I don't really do anything that tax the graphics but hoping to find a worthy task to fit into my workflow.

 

[cmaier]

To be fair, it’s an RISC CPU so it needs more cache to begin with.

Not really. And certainly not more data cache. Typically they need a little more instruction cache. And these are two different caches.