You are here :
Home
BenchIOS

Iphone and Ipad Benchmaking

Here are the last IOS iPhones and Android Phones/tablets CPU performances.

 

Apple A6 SOC, the best phone and tablet CPU performance ever !


The A6 is a dual core chip, but it manages to outperform quad cores in big integer manipulations. The 5000 digits of PI calculation shows clearly that Apple engineer made an extraordinary job.

The 5000 digits of Pi calculation show the real well architecture integration between the CPU and the memory. Pi calculation uses only integer number stored in huge array and the four operations  +-*/.

 

The numbers are in seconds.

Time to calculate 5000 digits of PI (seconds)

Only in the 4 simultaneous tasks operation the Galaxy SIII is slightly better : 0.618 s for SIII against 0.656 s for the iPhone 5 ! To outperform the iPhone 5 the program must written in native ARM code. Java code is from 2 to 3 times slower.

 Graphs :

Time to calculate 5000 digits of PI

Galaxy SIII performs better in native code, HTC One X performs better in Java code (Dalvik JIT interpreter).

 

 Time to calculate 2*5000 digits of PI

Shows a clear advantage to multiple cores CPU.

 

 time to calculate 4*5000 digits of PI

4*5000 digits of PI Shows the performance of the quadruple core architecture (S III, X one, Nexus 7)

 

 Those graphics show clearly that the iPhone 5 A6 processor performance are the best so far.

 

 Here Are all the results of the 5000 digits of PI computing. The PI Algorithm use a lot of integer and memory manipulations. Basicaly it is the core of the CPU that is evaluated in terms of computing speed.

Time in second. "," stand for "."

 Performance in second

Ratio compared to iPhone4 performance

Ratio performance compared to iPhone4

PI 1 task : One calculation of 5000 digits of PI.

PI 2 tasks : Two simultaneous processes running the same program (computation of 5000 digits of PI). Basically in a mono processor system it takes twice the time for two tasks. In a dual processor system it takes almost the same time for one task or for two tasks.

PI 4 tasks :  Four simultaneous processes.

Double : WGS to Lambert geographic transformation using only floating point numbers

FPU : WGS to Lambert transformation using the floating point hardware for trigonimetric functions like sin, cos, tan.

 PI one time

 

Two tasks

 

Four Tasks

 

Double computing

 

Floating Point Unit performance

 

Keywords : iPhone, Android, Bench, Benchmark, Apple A6, Tegra 3, iPhone 5, Galaxy S3, HTC one X. Dual Core CPU, Quad Core CPU.