What's new in this version (1.6.18.120610):

Current version: added LG-P990hN 2.3.4 - The fastest phone tested with this app. The only thing faster is a Galaxy Tab 10.1

Previous versions:
added Transformer Prime TF201 4.0.3 - numbers are a bit disappointing; maybe other running apps could be shutdown.

Added results for the HTC PH44100 with Android 2.3.5, also called the Kingdom, Hero 4G, and Evo Design 4G, depending on the carrier or locale

Description

This app provides valuable info to users:
- a comparison of one's device performance with other devices,
- an independent performance ranking of multiple Android devices,
- a spectrogram for acoustic analysis of live audio,
- free updates with additional models in new releases of the app.

This app does not automatically send any data to the developer; this is an option, the user must select the email function for any data to be sent to the developer. And when a user gives us permission, then we thank her or him here.

We would like to warmly thank all our contributors and specially those who have accepted that their names appear here:

Brannon Shipp
Frank Cemente

~~~~~~~~~~

This app can be used as a benchmark and as a spectrogram. As a benchmark, it compares the performance of your device with other Android devices. As a spectrogram, it analyzes sounds in real-time.

The application has an optional email function for returning your results to Simplecode. We compile anonymous averages for device models; these are published in the next releases of the app.

The benchmark includes an arithmetic component and a graphics component. The arithmetic component uses Fast Fourier Transform (FFT) calculations on live sounds from the microphone. The graphics component draws the spectrogram from the results of the FFT. The graphics component uses the Android 2D canvas library (not the OpenGL library).

The FFT calculations use parts of the NIST (National Institute of Standards) scimark2 program, as well as calculations developed by Simplecode. The NIST Java code, written by Bruce R. Miller, was inspired by the GSL (Gnu Scientific Library) FFT written in C by Brian Gough. The FFT calculations use 'double' primitives. Devices that have a floating point processor will tend to have much better results than those that do not have a floating point processor and are performing the floating point calculations in software.

The duration values are initially measured using System.nanoTime(), the most precise time value in Java. The Java specifications do not make any claim about the actual precision of the time measurements. The precision is dependent on the underlying hardware, virtual machine, and compiler.

During a benchmark run, the arithmetic and drawing durations, in nanoseconds, are added for each cycle. The results are shown in milliseconds after the benchmark run. Your best results are stored and used in the comparison chart. Repeating the benchmark a few times might improve your score.

For best results, other apps running during a benchmark should be kept to a minimum. In addition, the user should refrain from using other apps and from changing the orientation of the device during a benchmark run. The warm up period should ensure that the virtual machine is in a static state before the benchmark measurements start.

Small Icon © AguaSonic.com - made from cetacean sounds

App © Simplecode.com

Last comments from Android Market