Dolphin, the GameCube and Wii emulator - Forums

Hi there Dolphin forums.

This is my computer. I know the graphics card is nowhere close to top-notch, but I believe, albeit maybe incorrectly, that I should be getting better results than I actually am.

First things first, I have fully or almost fully updated drivers, and the games I'm talking about are ripped from my own and have not been scrubbed or altered with any other way.

I'm having issues running games at 2xNative, 1680x1050 Fullscreen, DSP LLE recommended, no VSync, 16xAF and deactivated AA, specifically Smash Bros games, Zelda games besides Wind Waker, and Mario Galaxy. Almost any game, to be honest.

On Skyward Sword I'm using D11, as suggested on the wiki, and I have less than 30fps on the town, boss battles, whenever there's a dialog going on. I only have 30 constant fps on certain areas, such as the area close to the temple, but not on Altarea. I have to admit that it looks stunning nevertheless.
Twilight Princess for Gamecube has similar issues.

Both Smash games stutter unless I use the OpenGL plugin, and it's strange because if I take the frame limit off, I'm getting +80fps. With it on auto, I'll have 60fps but every 10 seconds or so it drops to 40fps for about a second.

Super Mario Galaxy is unplayable. I'm getting 35-50fps no matter what plugin I'm using and the settings. I've tried 1x IR to no avail. Super Mario Sunshine is rather slow too, runs similar to TP Gamecube.

With Wind Waker I'm having no troubles whatsoever, whether I'm using D9, D11 or OpenGL, and it runs smoothly, at 30fps/60vfps with 3x IR, as is Four Swords Adventures.

Is this expected with my graphics card? Am I doing anything wrong? Any suggestions? I'll post any info that might help you identify the issue. I've seen on the wiki that computers, apparently far inferior to mine, are getting better results.

I'm using the latest Dolphin revision because with 3.5-134 x64 I was having issues with the rumble feature with my 360 pad. I'm also using a real, first party Wiimote+.

Please pardon my English, I'm not a native.

Your GPU is enough for 2-3x IR in most games, and your processor is pretty good. However, SMG in particular is a very CPU dependant game. If you had the i7 2600K, and a suitable motherboard, you'd have to overclock to 4.5 GHz to get fullspeed. It's expected that Wind Waker and the Smash Bros would run pretty well on that rig, as they aren't very demanding, but Skyward Sword may give you some trouble.

You mentioned using LLE. This requires a lot of processing power compared to HLE, so you should use it only for games which require it, in this case, just SMG. If you're using it for other games, then it may well be the source of your slowdown.

(01-09-2013, 08:31 AM)AnyOldName3 Wrote: [ -> ]Your GPU is enough for 2-3x IR in most games, and your processor is pretty good. However, SMG in particular is a very CPU dependant game. If you had the i7 2600K, and a suitable motherboard, you'd have to overclock to 4.5 GHz to get fullspeed. It's expected that Wind Waker and the Smash Bros would run pretty well on that rig, as they aren't very demanding, but Skyward Sword may give you some trouble.

You mentioned using LLE. This requires a lot of processing power compared to HLE, so you should use it only for games which require it, in this case, just SMG. If you're using it for other games, then it may well be the source of your slowdown.

Thanks for taking the time to answer. Yes, I was using LLE as the default configuration as I had the misconception that HLE would give poor audio output, but I was checking the option to lock it to a single core (or another similar option), in case it makes any difference. I didn't know I would not be able to play Super Mario Galaxy at decent speed, I expect SMG2 to be equally or more demanding. I guess I'll stick to play them on the Wii.

Should I be able to use any AA at all?

(01-09-2013, 08:56 AM)motdog Wrote: [ -> ]Should I be able to use any AA at all?

Depends on your IR, you should be able to do 2x AA and 2x IR

Often you'll get better picture quality from raising IR as opposed to AA, as they both do more or less the same thing, but AA gets downsampled twice before you see it as opposed to once.

I did getter somewhat better results with the HLE option. However, it sounds far worse: the sound difference in WW between using HLE and LLE is quite noticeable. Also, I didn't see an improvement in Skyward Sword, it got more stable in closed areas, but both in Altarea and especially during the final battle, I couldn't get it past 20fps at best, even at 1xNative.

I must have de/selected the appropiate option because now Melee and Sunshine run quite smoothly, even on 3xNative and with LLE on. However, Brawl is stuttering still, at 2xNative, and on Twilight Princess for the Gamecube, on some areas I can't get past 20fps either, like on the dessert. Maybe this one is a known issue?

http://imgur.com/a/3j20w


This is my current config for most games besides Skyward Sword, although it's normally on 2x. Is there anything wrong with it?

Thanks again.

You're texture cache shouldn't be set that safe unless your games crash without it. Put it to fast.

Quote:as they both do more or less the same thing, but AA gets downsampled twice before you see it as opposed to once.

Incorrect. SSAA applies a different downscaling filter that replaces the default, not a second downscaling filter. This is why you will notice that 1x + 4xSSAA for example looks similar to the sharpness/detail of 2x rather than 1x.

I thought that that was because 4^0.5 was 2, but I'll believe you and not press this further. I haven't done much testing with AA vs IR, so can't comment so much.

It's just a different way of writing it.

2x IR is:
2x scale on the x axis
2x scale on the y axis
4x total resolution

4x SSAA is:
2x scale on the x axis
2x scale on the y axis
4x total resolution

IR is written as the scale applied to each axis so the total resolution is the square of that scale.

SSAA is written as the total resolution so the scale is the square root of that.

IR really should be written as the total resolution in my opinion but the option was originally called "efb scale" so listing the scale made more sense at the time. It was renamed to "internal resolution" because people were confused about what it is for but they never changed the numbers to use a more appropriate measurement.