Controversy on the performance of Samsung smartphones

According to a survey of our colleagues Ars Technica, Galaxy Note 3 outperforms its CPU speeds, but only when it detects a test application, such as those used for comparisons.

The Galaxy Note 3 has amazing performance.

Samsung would he of "bulking up high-tech"? Surprised by the excellent performance of the smartphone Galaxy Note 3 confreres site Ars Technica have dissected the device to discover a very questionable behavior. When a standard application performance testing is started, the Snapdragon chip 800 of the camera switches to full speed: 2.3GHz on four cores.

The problem is that switching is programmed for this type of applications. Ars Technica has identified the Java file that causes the switch to "full speed" mode. In the code, references to test applications are clear: Geekbench, Quadrant, Antutu, Linpack, GFXBench etc.. These applications are often used to perform comparative tests.

For any other use - surfing the web, gaming, video, etc.. - The performance of the chip is reduced by approximately 20%. Renaming a test application referenced by the device also leads to the same result: the performance deteriorates. So this "full speed" mode would only serve to create beautiful notes in comparisons. Contacted by 01net, Samsung did not want to comment.

There was a history

This is not the first time that the manufacturer is pinning for this kind of practice. In July, the AnandTech site had noticed an unusual increase performance CPU / GPU on the Galaxy S4, again only for some test applications. This "anomaly" was also detected a few weeks earlier in a forum .

At the time, Samsung gave a somewhat convoluted explanation, saying that the performance was deliberately restrained for games, not to consume too much energy, but the maximum level was available on some applications like the browser, the camera or video player. But this statement appears to be incorrect, according to Ars Technica. By testing these different uses for the Galaxy S4, maximum performance could be achieved only sporadically on the camera.And continuously on test applications, of course ...