Dolphin, the GameCube and Wii emulator - Forums

The benchmark results are here. Thanks to Garteal for taking the time to do that.

Alright, the majority of the people who voted in the new benchmark thread have selected The Legend of Zelda: The Wind Waker as the game to test with, so here we go.


The Legend of Zelda: The Wind Waker CPU Benchmark

Test Build

• We're going to test with the Official Dolphin 3.0 release.
  
• Download Windows x64.
  
• Download Windows x86.
  
• If you want to test with Mac or Linux, you'll have to use your own builds. Just use these settings and use this memory card file.
  
  

Test Area

• There is a Zelda Wind Waker NTSC memory card included with both builds above.
  
• If you're using a game region other than NTSC, you'll have to convert the save yourself if you want to participate.
  
• There are 3 saves on the memory card, this means that there are 3 separate areas to test.
  
• Load each memory card save on "Memory Card A" (1, 2 and 3) and do not move your character. Record your game speed, preferably in a window with the title bar displaying fps, vps and speed as I did below.
  
• You should also manually type out what the FPS was above your image as eventually, image hosting sites tend to remove some images.
  

Settings

• The settings are preset, you don't have to change anything.
  

Reminder

• List your PC specifications.
  
• Post a screenshot of your CPU speed while running this test, use a program called "Core Temp" or "Real Temp" to do this.
  
• Let us know which game region you're using.
  
• Let us know which Operating System you're using and if you're using x64 or x86.
  
• Don't change any settings.
  

My results

Asus Laptop: K53TA
OS: Windows 7 Home Premium, 64-Bit - SP1
CPU: AMD Llano A6-3400M, Quad-Core, 2.3GHz (Overclocked)
GPU: AMD Radeon HD6650M, 1GB GDDR3 (Catalyst 12.4)
RAM: Samsung 4GB DDR3-1333

Dolphin 3.0 - 64-bit. NTSC.

The only options that were changed were: Forced 16:9 and I used the Widescreen Hack. My native screen window was 1366x768.

This is with the first save on "Memory card A".

FPS/VPS/speed
32/64/106



Second "Memory Card A" save.

FPS/VPS/speed
26/54/90



Third "Memory Card A" save.

FPS/VPS/speed
40/81/135

For what it's worth, I wanted to include some other areas, but that would mean that testers would have to do some walking and repositioning. This way, just load the save and you're where you need to be.

I have two computers to test, so here is the first one:

MacBook Pro Early 2011
i7 2720QM @2.2 GHz, turbos (and is throttled to) 2.8 GHz while running Dolphin.
AMD Radeon HD 6750m
8 GB 1333 MHz RAM
Windows 7 x64 SP1

And here are the results:

1st save file:



2nd save file:



3rd save file:



And here is the other computer:

Desktop I built
i5 3570K @ 4.4 GHz
NVIDIA GTX 550 Ti (EVGA Superclocked)
16 GB 1600 MHz RAM

Settings were left as-is

1st save file:



2nd save file:



3rd save file:



Dolphin froze once because the FPS went too high. I entered the third save file only to be met by a black screen and 1551FPS *_*

Processor/CPU: AMD 955 OC'd to 3.434
Video Card/GPU: ATi 5830 OC'd to 850/1100
Memory/RAM: 4 GB @ 1616MHz
OS: Windows 7 64-bit

I used 64-bit dolphin, WW was the NTSC version, and only thing changed was the controller settings :3

Save 1: 45 FPS / 90 VPS
Save 2: 39 FPS / 79 VPS
Save 3: 60 FPS / 120 VPS

My results seemed to have been affected by the higher IR after all, I set it to 1x and retested.

I was going to recommend that you get rid of that part about allowing changes to the IR under the assumption that it won't impact performance but I refreshed and see that you already took care of that.

So in that case I have only one other tiny thing to recommend.

Axxer Wrote:i7 2720QM @2.2 GHz, turbos (and is throttled to) 2.8 GHz while running Dolphin.

See how he listed the speed that it was running at with turboboost active and not just the base clock rate? That is must have data for anyone with a cpu that supports turboboost that hasn't been manually OC (since manual OC overrides turboboost). It is crucial that we know the exact clock rate that the cpu is running at or the results will be inaccurate. This can be done a number of ways, cpuz is one way. I know it's a lot to ask since it requires 3rd party software but it is kind of a deal breaker if we want accurate results. For example two people may have the same cpu and get different results because one turboed higher than the other due to better cooling.

Turbocore (AMD equivalent such as yourself) users should be exempt from this rule since it is impossible (at the moment) to report the correct clock rate using software due to the way turbocore works. Turbocore increases the clock rate in short bursts ranging from a few nanoseconds to a few milliseconds, which makes it impossible to accurately gauge with software.

Aspect Ratio : Stretch to Window , Render to main window , Wide Screen Hack
1x IR doesn't make the game run faster on my laptop

1st save file: - 2x IR , noAA , 16x AF


2nd save file: - 3x IR , noAA , 16x AF


3rd save file: - 2x IR , noAA , 16x AF


New Benchmark result : http://forums.dolphin-emu.org/Thread-the...rk?page=11

Specs : i5 480M @ 2.9GHz - Nvidia GT 540M (Stock)

Please keep it at 1x even if you think it won't effect performance. I'll have my results up either tonight or tomorrow afternoon (more likely the latter since I'm lazy).

(09-20-2012, 12:04 PM)NaturalViolence Wrote: [ -> ]Please keep it at 1x even if you think it won't effect performance. I'll have my results up either tonight or tomorrow afternoon (more likely the latter since I'm lazy).

Hmm. I'll have to dig up my copy of WW.
Maybe in a few days.

(09-20-2012, 11:47 AM)NaturalViolence Wrote: [ -> ]I was going to recommend that you get rid of that part about allowing changes to the IR under the assumption that it won't impact performance but I refreshed and see that you already took care of that.

So in that case I have only one other tiny thing to recommend.

Axxer Wrote:i7 2720QM @2.2 GHz, turbos (and is throttled to) 2.8 GHz while running Dolphin.

See how he listed the speed that it was running at with turboboost active and not just the base clock rate? That is must have data for anyone with a cpu that supports turboboost that hasn't been manually OC (since manual OC overrides turboboost).

I was just used as an example of good practice by NaturalViolence? I have been blessed by Nayru this day.

Anyway, I'll go back and edit the post for my MacBook Pro to reflect using 1x IR later when I have the time.