Samsung’s Throttlegate spills over to the Tab S8
As the Game Optimization Service is on a wide variety of Samsung phones and tablets, Throttlegate continues. There’s also really very little in terms of interesting news in the Android sector so I’d expect to see this for every Samsung device down the line including discontinued devices. The Tab S8 is the latest listed, same story as all the rest with the Game Optimization Service.
TL;DR – lot of opinion on a Monday morning about benchmarks.
The code slows down CPU and GPU performance for thermal and battery reasons. It was not configurable. It is now configurable. Patches are rolling out. Nothing says giant evil conspiracy like selling a brand new phone and making it not perform well.
This put benchmark companies in a pretty bad light, which is why I think this story keeps going. Point at Samsung for the benchmark failing like all benchmarks do.
Now, if Samsung had sold the S22 and not had game optimization on it and it was on the older devices and started cranking down performance, there you’ve got a good story, another Apple “battery management” story where old devices were being slowed to a crawl. A uniform throttle app across all devices, not something I’m seeing out of Dr. Evil’s playbook.
I guess if you’re a benchmark company, and Samsung didn’t similarly throttle your benchmark you might have cause to rise in anger because you’re not able to produce a representative reflection of the actual performance of the device. “TestX claims 100FPS in Game Y” – well, no, it got 100FPS in your benchmark. Your benchmark is not the game and as such it’s not a representation of how the game actually responds.
If you’re going by benchmarks and these benchmarking software orgs as actual indicators of performance, you have some other problems to contend with though. This is mostly an embarrassment for benchmark software companies because they’re being shown to have the same issues they’ve always had: they’re not testing the game performance, they’re testing what the benchmark company thinks the game performance is based on some test they devised.
Any optimization software, battery saving, etc throws this completely off. Even the position of the phone during testing can change performance due to thermal changes/throttling, power needed for connection, relative humidity, time of day, bears in the neighborhood, etc.
Until the benchmarks are network off, cold room, same placement, same time from boot, fresh image, actual robot playing a game (using the real inputs,) and counting FPS instead of a faux game being executed, you’re not going to have real numbers. Benchmarks are for estimates and people who think benchmarks are performance. Oh they work, but yeah.
Geekbench is currently banning Samsung products that were affected from ever having a GB score it looks like because… well, I guess GB thinks they look bad because Samsung, rather than because they didn’t realize that things in the GAME category run under game optimization services, and if you’re testing a game you might need to be there to have a representative experience.
[Android Police]
