S9/S9+ Exynos Info

[Nemesis]

Hi to all!

I just installed latest build on my phone (with Pie) and playing Mario Sunshine.

It works at 30fps without override...unless I press a button. If I don't press a button it falls at 15-20 fps...

Just tried also Fzero GX and it runs at 40fps, is there some nice config for this game?

[Helios]

Yep. Samsung phones do that.

Their CPU clock speed management is fairly conservative until it detects user input, then it ramps up clock speeds to keep things responsive.

[TMCThomas]

I Must say that the build of weihuoya for me "fixes" this issue. Especially apparent in games like wind waker. Fps stays fairly stable even though I don't touch anything. I can back this up with evidence of someone doesn't believe me.

[JMC47]

If you want to point out what change was done to do this, that'd be better. The fork doesn't have well labeled commits or changes, so it's really hard to find any individual thing it does. Like with other changes people have found in it, it'd be more worthwhile to actually figure out what it's doing to trick the governor.

[Nemesis]

Just to add an Info. If I use a gamepad, the analog stick movement doesn't count as a pressed button, so speed will be the same. To gain speed you have to touch a virtual button or a physical pad button but not the sticks.

[TMCThomas]

honestly, I don't know. I think @weihuoya would be the only to know that.

[nonexist]

nah this only happend on samsung exynos phones. samsung does very strange things with his CPU

[Guilherme]

Samsung CPU governor:

Code:

if(benchmark) {
high_performance_mode = true;
}
else {
save_battery_mode = true;
}

[nonexist]

in my experience with samsung exynos devices even forcing perfomance mode they are very slow for emulator and the only way to fix it is to use custom ROM/kernel

but now I'm more than happy with snapdragon phones.

[Guilherme]

Well, I was just joking...

With my previous Exynos phone, the only improvement I had with custom ROM/kernel was the thermal throttling which was less agressive.