home
A few days ago came the news that the Galaxy S4 smartphone from Samsung has cheated a number of popular benchmark application. This smart phone is suppose GPU clockspeed increase in applications, such as the desire to push AnTuTu and GLBenchmark scores which produced.
Galaxy S4 allegedly using a special application to the string "Benchmark Booster" which automatically detects the user running the application benchmarks, the GPU clockspeed boost above the normal value.
Related to this, the Samsung has issued an official rebuttal, published in the company's blog. The Korean manufacturer said that a higher clockspeed is also applicable for other software outside of benchmarks that are run by the Galaxy S4.
"The maximum GPU frequency of 533 MHz is used for running applications in full-screen, such as S Browser, Gallery, Camera, Video Player, and certain benchmark applications that require high performance," Samsung wrote in a statement, as quoted by AnandTech.
In other words, Samsung refused to say just boost the speed of the Galaxy S4 to produce a benchmark in value above the actual capabilities of the device, in daily use. Some games are called only allowed to reach 480 MHz frequency for stability reasons.
"The frequency of maximum of GPU-owned Galaxy S4 deliberately varied to provide optimal user experience, and are not intended for a particular benchmark," added Samsung.
Up and down
But is it in fact according to claim the Samsung above? As the technology site AnandTech investigate it deeply re-do the test re-adjusted by the manufacturer Galaxy S4 statement.
The result was slightly different from that described by Samsung. The gaming software is running at a frequency of 480 MHz, but some software natively for full-screen (S Browser, Gallery, Video Player) apparently stuck at 266 MHz GPU speed, not 533 MHz as Samsung claims. The native applications basically does not require a high performance GPU.
One exception contained in the camera application that occasionally reach speeds of 533 MHz GPU when the user is actively experimenting camera filters (especially the kind of "rugged", "oil pastel", and "fish eye"), but it will not last at that frequency is constant such as application-specific benchmark applications.
Based on the graph created by AnandTech below, it appears that the Galaxy S4 GPU speeds up and down in the camera app, not consistently be on top like AnTuTu benchmark and GFXBench.
Why does this happen? Galaxy S4 allegedly opened limits heat (thermal limit) for a number of benchmark applications so as to maintain the speed of 533 MHz GPU. The camera software which keep getting the restriction still not able to maintain the frequency of the GPU in the upper limit.
Increased frequency of GPU here might just boost benchmark scores by 10 percent, but the results have been very good possibility considering the hot conditions that limit the Galaxy S4.
Still "cheating"?
So, does this mean that Samsung remains valid "cheating"? A reader commented that the site AnandTech benchmarks should be run in the same condition as daily use (for example, when running the game) so that it can obtain a fair and appropriate value of the real situation.
AnandTech suggest Samsung to give users the option to run mode "turbo" is as desired. Another option is to remove it entirely so that eliminating the optimization of confusion about this.
Home
