When Samsung’s Galaxy S4 hit the market and reviewers hands, it was quite impressive overall. One area in particular was benchmarks, and now we’re starting to figure out exactly why. Don’t get us wrong, the Galaxy S4 is a splendid device and possibly the best smartphone on the market, but new reports are surfacing that they specifically tuned the GS4 to score higher than usual in popular Android benchmarking apps.

The always thorough guys over at AnandTech have done some investigating, and revealed that benchmark apps received extra attention from Samsung. And it goes so far as to basically show that benchmarks even received more power from all those processors than regular apps, as a way to inflate the benchmark scores.

This mainly breaks down the Exynos 5 Octa chip found in select Galaxy S4, as the US and other regions all have Qualcomm’s chip. The Exynos 5 Octa and Snapdragon 600 reserved higher speeds only for benchmarks, and real world usage didn’t receive the same. It’s a pretty interesting and detailed report, and makes us question benchmarks as a whole. For example the GPU would crank up to 533 MHz for benchmarks, and run at a speed no users apps or games will ever be able to enjoy. Stuck at 480 MHz for regular use. Which of course makes it look better than it really is.

As we all know, benchmarks are only numbers and don’t represent real-world usage. For all of our readers, you know every review we do we specifically show benchmarks, but follow that up with real world usage. As benchmarks are only a small part of the story. They can’t fully be trusted, have always been a grey area, and are more for bragging rights than anything.

To be honest though, the PC world has been doing this for years with NVIDIA and ATI. It’s nothing new, as manufacturers are always striving to offer the best results to show off their impressive products. Two wrongs don’t make a right, and we’d love some transparency here, but either way the Galaxy S4 is still a great device. Even if this was a little deceiving.

We’ve reached out to Samsung, who had no comment at this time. We’ll update if we hear more.


  1. What sites publishing this news don’t know is that Samsung didn’t promote the number of MHz in its GPU, it didn’t even mention the model. Those test suites run at SAME frequency than the GPU (PowerVR SGX 544MP3, 533MHz). How is that translated to “tweaked to achieve unrealistic benchmark”? What I don’t know is the decision behind the “topped at 480 MHz”, but that is a different story.

    • What the hell are you talking about? This IS a big deal. Samsung is not advertising that speed, its true. But that is not the issue being questioned here. All the benchmark results are artificially inflated and dont provide an accurate measure of what the terminal is capable of. This, on top of that, was done willfully as stated by the code that Anandtech discovered.

  2. I’m pretty sure Usain Bolt doesn’t cover every 100 yards on the ground
    at sub 10 second speeds. I’m pretty sure he reserves those performances
    for track meets, especially when it really matters – like, you know,
    Olympic or World Championship finals races. I don’t think the fact that
    he reserves those speeds for those moments invalidates his claim – which
    is widely recognized – to be the fastest fastest athlete at the
    distance in our time. Or do you think those cars on our roads cannot
    achieve 200mph just because most of the time they move along at less
    than 65mph? Duh!!!

  3. Samsung proves that, they are world biggest frauds, by selling Galaxy S4 with false promises. just 9 GB instead of 16 GB as Samsung advertise and claimed, and not able to copy or transfer Apps, games or photos to Micro SD and wise verse. Samsung you big cheat.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.