Dolphin, the GameCube and Wii emulator - Forums

(10-23-2011, 01:46 PM)TheDarkFenrir Wrote: [ -> ]anyways, I meant to ask, APU's are out of the question huh? (even Overclocked?)
I heard a APU-DOES HAVE intergrated graphics BUT its supposed to be on par with a 64xx AMD GFX card.

Laptops like the one in my signature use an APU and it does work with Dolphin. The video card is on par with an HD6570 desktop video card, it's not the greatest video card in the world but it is quite good for a laptop. I won't tell you that all games run well, but there is a good amount that do. Wii games are out of the question, though.

Lol, I see.
I couldn't get Wii games running on this laptop at all...
Well let me take that back..Brawl managed a good 17FPS...but thats about it..
Most gamecube games fail to run and i have a few that I haven't tried on this thing.
Anyways I was asking about APU's because I am building a desktop and I haven't decided if I should just go with the APU or phenom II x4 840+AMD5450 GPU

And to be honest..its time for me to upgrade >_> this thing cant even run pc games without failing doing that.

My A6-3400M APU is very close in performance to the Phenom II 840, but my HD6650 is much better than an HD5450. Either way, you're going to need something more powerful unless you want to be limited in what you can play.

How about the XFX Radeon HD 4670?
That way I can retire this old celeron

The HD4670 is close in performance to what I have, but you should do better than that.

Quote:The video card is on par with an HD6570 desktop video card

Not quite. The absolute fastest llano integrated gpu is the 6550D which as you can guess by the number is inbetween the performance of a 6450 and 6570 desktop graphics card, still good enough for dolphin though.

And just to be clear to the OP (I know that you know this SS) the reason that the APUs won't do so well in wii games is because of the poor cpu performance not the integrated GPU. Getting a graphics card would not help unless you are planning to run dolphin at a high internal resolution.

Quote:My A6-3400M APU is very close in performance to the Phenom II 840

I highly doubt that. The phenom II 840 has the same core architecture (K10-II) but with a larger cache and a 50% higher clock rate (and that's assuming turbo core reaches its maximum clock rate which won't happen if 3 or more cores are active).

My GPU. Not including a small boost for crossfire.



HD6570 video card.



Close enough for me.

Yes but you're looking at your discrete GPU. The A6-3400M APU isn't always paired with a discrete GPU in laptops (in your case it is).

I guess you were talking about your discrete GPU when you said "video card" so I misunderstood.

@TheDarkFenrir
Both the 5450 and 4670 are greatly inferior to the GPU in llano and are too weak for dolphin.

@SS
Based on those specs the texture fillrate, shader throughput, and video memory bandwidth are all roughly the same in both your 6650M discrete GPU and a 6570 graphics card. However your discrete GPU has half the pixel fillrate of a 6570 (due to having 8 ROPs instead of 16), which I think is rather odd since ROPs don't take up a lot of die space. This means you should probably stay away from AA. I am curious how the pixel fillrate effects dolphin. Reminds me to organize a test for you one of these days.

For what it's worth, the passmark CPU look-up shows that the Phenom II 840 is slightly better than my A6-3400M, but not by much. It's not a Dolphin bench, but that's all we have to go on right now.

Passmark has never been very reliable in my experience. Common logic says the phenom II 840 would be about 50% faster in single/dual threaded applications and twice as fast in heavily multithreaded applications (3/4 threads with good load balancing) since it's essentially the same cpu but clocked higher (and fabricated on 45nm with no integrated graphics processor).

Something has to be wrong with that benchmark to get those kinds of results. Every benchmark I've ever looked at on anandtech/tomshardware/lenzfire/guru3d/etc. shows the desktop llanos being inbetween an athlon II and phenom II in performance per clock per core. In order to get those passmark results the performance per clock of of the phenom II 840 would have to be half of the llano since it runs at double the clock rate in 4 thread applications and 50% higher in 1/2 thread applications. This means that the laptop llanos would have to be massively faster in performance per clock than there desktop equivalents, which doesn't make any sense because it's the exact same chip.

The only thing I can think of which could logically explain those results in a properly conducted test (assuming it was properly conducted which I doubt) is if passmark was a totally memory bound application. If passmark is totally bottlenecked by memory performance that would explain the two chips performing similarly but I can assure you dolphin is not memory bound. And considering the rest of the passmark results are not consistant with memory performance it is far more likely that the person conducting the benchmarks screwed something up.