IPhone 7

>iPhone 7
>4 CPU cores
>multicore performance of the 4 core is less than 2x the single core

Anyone else think that doesn't seem right? Is the apple CPU a dud?

I know you don't get perfect multicore scaling and diminishing returns when you add more cores, but this seems worse than usual...

Other urls found in this thread:

anandtech.com/show/9837/snapdragon-820-preview
geekbench.com/doc/geekbench4-cpu-workloads.pdf
twitter.com/NSFWRedditVideo

>can't run Linux

Its big.LITTLE.
Only two big cores are used in intensive workloads.
Multicore scaling is never 100%.

Still, you'd expect more than double single core with that config. The SD820 is a quad core with big.little and it gets over 2x single core.

>They finally caught up with Mediateks 2015 CPUs

HAHAHAHAHAHAHAHA

Also it's probably thermal throttling.

The Snapdragon 820 is not big.LITTLE. It is 4 of Qualcomm's Kryo cores dividing into two clusters running at two different clock speeds with different L2 sizes. They're the exact same core arch.
anandtech.com/show/9837/snapdragon-820-preview

Read the review.

>2 clusters
>one faster than the other
>not big.little
Lol okay, Fast.Slow then, whatever works for you.

big.LITTLE has a specific meaning.
ARM designs distinct stock IP for cores that operate in different power ranges.
Taking one core and having two default clocks rates is not big.LITTLE. The 820 can use all 4 Kryo cores at once if needed as thermal margin and power allow.

So¿ You want your iPhone heating up on single core 100% while I have 4 cores, throttled down doing the same amount of work for less.

>geekbench

How can Android even compete?

Run on iPhone7 hardware.

I have an iPhone 5.

It runs iOS 10.

Everything is snappy and I've never had an issue with performance.

Don't bother using reason. iToddlers will believe anything Applel tells them to believe.

Not an argument.

2 big cores, 2 small cores. Only one set of cores runs at the same time. Hence it is just a dual-core for the purposes of the benchmark. The scaling is then reasonable.

Geekbench is bad for comparing across architecture (ARM vs x86), but Android phones and iPhones use the same architecture (ARM), so it is suitable for comparisons.

How is that suprising? The 12 core mac pro runs a similar IPC lower frequency intel core on a worse manufacturing process, it is completely normal for the iphone 7 to beat it in single threaded short runtime benchmarks.

>Android phones and iPhones use the same architecture

>iToddler level technology literacy

>Android phones and iPhones use the same architecture (ARM)

Yeah. That's why x86 AMD and x86 Intel CPUs perform identically in both benchmarks and real world performance at the same clockspeeds.

Why do mactard morons get the false impression they or their fruity toddler toys belong on a technology board?

I think Geekbench is skewed in some manner towards Apple devices, honestly.

There's no argument to be made that iOS itself is more efficient than Android, anyone trying to argue otherwise is just a fucking idiot: iOS runs bare metal on the hardware, Android still sits on a virtual machine because of how it's designed so obviously code running on iOS is going to run faster.

But if Apple is professing that their SoCs are just that much faster in terms of processing performance at such lower clockspeed then something is fucking wrong someplace or someone is fucking cheating.

Personally I'd lean towards the cheating aspect. If it can be done on desktops it can be done on mobile platforms just as easily and probably more so because it's a narrow profile platform meaning not a whole lot of configuration options like a desktop PC has which is nearly infinite.

Much like how tech journalism favors apple, I get the feeling geekbench slightly favors apple

HOLY SHIT #BTFO

He was referring to instruction set architecture I think, in which case they do run the same

>instruction set
see

>Yeah. That's why x86 AMD and x86 Intel CPUs perform identically in both benchmarks and real world performance at the same clockspeeds.

When did I imply that because they are both ARM they should perform the same? Obviously they don't. Hence the differing scores.

Geekbench is not suitable for comparing x86 CPUs to ARM CPUs. It is suitable for comparing ARM to ARM. That Apple btfo's the competition despite half the cores and same or lower clockspeed (not sure what it's at now) is testament to their massive IPC advantage.

Lets stop with the geekbench is not good for comparing different oses/different architectures meme. Geekbench 4 runs the same exact workload on all systems and architectures, using code optimized for each architecture, and it uses neon vector extensions on arm and all sse and avx extensions on x86 wherever the code can take advantage of vectorization.
If you don't believe me you can check www.geekbench.com/doc/geekbench-4-cpu-workloads.pdf and maybe come back with a real argument instead of random shitposting like
> I don't believe apple can beat intel so geekbench must be unfair

*geekbench.com/doc/geekbench4-cpu-workloads.pdf

Geekbench is bad for comparing even the same architecture.

You can only really use it to compare the same CPU family.

what are the other 2 little cores used for then?

Since Android and iOS are different OSes (that supposedly work on the same architecture) it is entirely possible to skew the benchmarks in favor of one over the other.

It's been done forever in "regular" computing benchmarks and it most certainly can be done for mobile devices as well.

"Some things are true whether you believe them or not..."

>tfw mt6795 helio x10 user
Posted from my Xiaomi Redmi Note 2 using Clover

Low load, long term processing.

When we are talking about natively compiled code that runs >99% cpu time and doesn't contain many system calls os differences are completely irrelevant as the os is going to be executing for less than 1% of the duration of the benchmark. The cheating in regular computing benchmarks you are taliing about hasn't got much to do with the difference in oses and more with specific compiler optimizations/cherry picking code paths that run better on a certain cpu, and if you bothered to look at the document you'd have seen see gb4 doesn't do any of that. Further proof is the a9x was already very close to core m in specInt 2006, which is an industry standard benchmark for integer performance and has been designed with the collaboration of many different cpu companies. What I find ridicolous is that people are still surprised that apple has almost caught up to intel big cores in ipc and beats them in efficiency while steamrolling other arm designs when apple is a 10 times bigger company than intel or arm, and has been hiring the best cpu engineers from intel, arm and amd for basically the last 10 years.

>apple is a 10 times bigger company than intel
Maybe if you count employees, and include all their store employees.

The thing is people link geekbench 3 which is a piece of shit, the newer version is OK.

What makes the newer one okay?

see

Do all programs use CPU instruction sets like SSE or AVX?

How can android run on x86 and android if this is the case? And if not then wouldn't that give the processor with more instruction sets an unfair advantage?

>calls people Itoddlers
>actually giving a shit about a fucking mobile phones cpu.

big.LITTLE is two clusters with compatible architectures, but they aren't necessarily identical clusters. The point is that you can have a lower power, simpler architecture for menial tasks and switch to the cluster with more power-hungry, more complex cores for more intensive tasks

No, only programs that use native code and are compiled with the necessary optimization flags.
Most android programs are written in java, those that use native code parts like geekbench have many different binary parts for each instruction set architecture. It is not an unfair advantage for x86 and its many extensions because the purpose of geekbench is to measure the maximum performance the cpu can achieve, not the performance of the badly written, unoptimized code that most programs use.

>2016
>super advanced phone
>STILL FUCKING CANT JUST COPY AND PASTE A FEW PICTURES TO PC VIA USB CONNECTION

First started using an iPhone about half a year ago and got a 6S. I plug my phone in and transfer photos all the time. What do you mean?

im used to android, shits so simple and it works.
decided to try out iphone and everything is so fucking complicated and asks me to log in to like everywhere so i can get my social experience.

That's not surprising since A10 has much higher IPC than even Skylake/Kaby Lake, let alone Ivy Bridge.

>fruity toddler toys
I sense a lot of juvenile jealousy.

are you the same guy that posted that he's gotten an iPhone 4S to replace his HTC M9?

This. I went from an iPhone 5 to a redmi 4 pro and holy shit this is a massive improvement

It runs another Unix-based OS.