Quote:Like I said earlier there really is no correlation between modern PC games and dolphin. A High end gaming pc will probably also be able to play dolphin but you can build a really good gaming system that won't be very good for most games on dolphin.
Indeed. The examples he listed of PC games were all GPU heavy.
Quote:Trinity is piledriver based so there are only two modules and it looks like dolphin is treating it like its a dual core processor.
No. I asked him to try lock threads to cores for a reason. Thankfully he left task manager open in his screenshots. You can clearly see that dolphin is using logical cores 1 and 3 as it should.
Quote:Is there anyone else on the forums with bulldozer or piledriver chips that have discussed the results they get? I haven't been here long and have only seen people talk about how bulldozer/piledriver should theoretically perform.
We don't have exact numbers from any of them, but yes. They generally perform similarly to high end phenom IIs from what we can tell.
Quote: His wind waker benchmarks aren't that far behind what I posted for a phenom ii x2 550, in fact his second save fps bests it. Maybe this is the best trinity can do on Dolphin?
I highly doubt that. Especially his SMG and NSMB results. When my Q6600 at 3.2GHz is getting 3.5 times the vps something is clearly wrong.
I'll check it out on my lads steambox setup which is similar hardware with just less RAM (8GB)
A10-5700 ITX.
Im not expecting it to be any better, just to confirm that the A10 isnt suitable.
Hello btw im manic-miner
and long term user of dophin (which is awesome)
The only suggestion I could possibly have for you is attempt to use some of the multi-threaded settings. The people around here probably won't endorse this, but I personally received a decent speed boost (10 or so frames) when running under single-core mode with an i5 (just for testing reasons) when I had LLE on it's own thread on a less intensive game. Also, try DirectSound, I've heard it runs on it's own thread which may or may not be true. These are just things to try out and I don't guarantee their accuracy. Might want to try using Open MP as well, which, because of Piledriver's architecture, may or may not give any improvements.
AKNova7 Wrote:The people around here probably won't endorse this, but I personally received a decent speed boost (10 or so frames) when running under single-core mode with an i5 (just for testing reasons)
It can happen. DC mode is generally faster but not always. It introduces some extra overhead.
AKNova7 Wrote:when I had LLE on it's own thread on a less intensive game.
LLE on thread does reduce the performance hit of LLE and is generally recommended on quad core cpus (and higher). However he is using HLE which is even faster.
AKNova7 Wrote:Also, try DirectSound, I've heard it runs on it's own thread which may or may not be true.
DirectSound is just a backend. It should not impact performance at all. And no it does not run in a separate thread as far as I know.
AKNova7 Wrote:These are just things to try out and I don't guarantee their accuracy.
None of the things you have listed should effect accuracy. LLE on thread can cause some games to hang (mainly games that use zelda ucodes) but other than that everything else is fine.
AKNova7 Wrote:Might want to try using Open MP as well, which, because of Piledriver's architecture, may or may not give any improvements.
Unless the game is doing a lot of texture decoding (the games that he has tried so far don't) it won't effect his performance.
Nothing is wrong with his settings. Something is wrong with his system that is causing him to receive only a small fraction of the performance that he should be getting.
I have an A10-5800K and have been meaning to bench it. I will try and do so soon and see if the results are slightly better (like you would expect) or a lot better (meaning something is wrong with yours somewhere).
I also have the A-10 5800k everything is playable at full speed. Some games I can run at 2.5x and some at 2x. I never use AA because it's kind of pointless to me. A couple weeks ago, my wife and I replayed through Mario Galaxy 2 and we had it on 2.5x, 0AA, 16AF. We never experienced any slowdown or hiccups.My little brother has the A-10 5700k. Everything performs pretty much identical on his machine.
Quote:We never experienced any slowdown or hiccups
Mario Galaxy 2 need LLE backend (dsp_coef.bin & dsp_rom.bin) . You're using HLE so that does not count
Even with HLE I have trouble believing there were no slowdowns at all. Also there are no amd cpus with the model number A10-5700K.
I have an AMD Phenom II 3.2ghz Quad Core which is a few Generations behind AMD A10-5700 and I can run New Super Mario at Full Speed, as well as Super Smash Bros. (not tried Galaxy).