Dolphin, the GameCube and Wii emulator - Forums

Hello, I have very recently encountered Dolphin, and after a lot of annoying newbie problems, I've managed to get it to work very well. (I'm on XP, though, and it appears some games glitch on XP whilst they don't on Windows 7... Oh well.)

Something has come to my attention after some time, though. My games lag. A little, but sufficiently that I found it odd. Without the dual core support, they would never attain full speed, but with it, I can run with high settings. The thing is I don't have two cores. I have eight. And I did check the CPU usage of dolphin when my games aren't at full speed. Dolphin only takes 25% (Which is logical, as I get "Dual Core" is for 2 cores). And it uses the two cores fully. But the 6 other cores aren't doing anything at all, so I am feeling it's power gone to waste.

Is there any way that Dolphin can make use of those extra cores? Thank you.

No, there isn't anything you can do currently.

Last time i checked dolphin only supports 2 processors, so sorry

You have 4 cores actually. The other 4 'cores' are virtual threads. And while it is disappointing dolphin and pcsx2 don't make full use of quad cores yet, you are seeing a performance boost regardless. I went into the BIOS once and disabled 2 cores on my i7 only to see fps drop by 20%

As far as i'm aware, OpenMP is supposed to give a speedup depending on the amount of cores that you have in your Processor. Try enabling that.

OpenMP only speeds up texture decoding, which won't affect gamespeed in any way on the vast majority of setups in the vast majority of games. There are a few games like mario kart wii that shows a difference but other than that don't expect to see any difference.

Since so many people are talking about this, maybe the devs should work on speeding up LLE on thread and OpenMP since so many processors now have more than two cores. Right now those options rarely provide speedups.

Many people have been talking about this for the last 2 years. But no matter how many people want it that doesn't help developers do the impossible. Skid doesn't think LLE on thread is going to get any faster, it's pretty much as good as it can possibly be right now. And openMP just improves texture decoding speed, as I discussed above.

I think the devs should have to rewrite the whole emulator from scratch if they wanted it to make use of all available cores... And it sure would be difficult.

Well, can't even PhysX/CUDA for counting lots of things [mainly precision floating points numbers] be implemented? Surely it would help increase performance a lot.

btw. does that OpenCL checkbox actually help or not? I'm starting to find these help tips on checkboxes kinda unreliable, so that's why I'm asking.