Dolphin, the GameCube and Wii emulator - Forums

I ran the benchmark on my father's workstation laptop.

Hardware configuration used:
CPU: Core i7-2640M @ 2.8-3.3GHz (32nm Sandy Bridge, dual core + hyperthreading, 2x256KB L2 cache, 4MB L3 cache)
GPU: Nvidia Quadro 1000M
OS: Windows 7 Ultimate 64bit SP1

NTSC-U Wind Waker

Summery of the performance:
Save slot 1: 181%
Save slot 2: 148%
Save slot 3: 225%

Save slot 1


Save slot 2


Save slot 3

My brother wanted me to see if his CPU could unlock and overclock so I tested it in my system since his cheap MB could do neither. I thought Why not run this benchmark while I'm at it. The only thing changed between this test and my last one is the CPU.


First no tweaking just stock cpu.
OS: Windows 7 home premium x64
CPU: Phenom II x2 550 (Stock 3.1Ghz)
GPU: Gigabyte GTX 660 2GB GDDR5
RAM: OCZ Gold 8GB DDR2
Dolphin: 64bit From this thread only options changed where the ones Starscream changed.
GAME: NTSC


I resized the images this time so I will type out the FPS incase it's hard to read.
First Save. 40 FPS



Second save. 31 FPS


Third Save. 49 FPS



Now you may be thinking this is the practically the same as the 1100t and in fact when I first opened the first save I was pretty disappointed. However, though a single pic can't show this, the framerate was not nearly as stable. Though the FPS In the pics was the most prevalent there would be short dips in fps by as much as 10 - 15 frames. Where as the 1100t was pretty rock solid with there only being about a 1-2 fps change in both directions.

Now for the Same tests with the CPU unlocked to a quad core.
OS: Windows 7 home premium x64
CPU: Phenom II x4 B50 (Stock 3.1Ghz, same CPU as before but Unlocked)
GPU: Gigabyte GTX 660 2GB GDDR5
RAM: OCZ Gold 8GB DDR2
Dolphin: 64bit From this thread only options changed where the ones Starscream changed.
GAME: NTSC



First Save. 42 FPS


Second Save. 35 FPS


Third Save. 57 FPS


Again what can't be seen in the pics is how much the FPS changes. Unlocking to a quad brings back the fps stability that the 1100t had. No more large drops in fps. The fps was generally within 2 of what the pic shows. The only exception is the third save. It was consistently between 50 and 57 but never really on anything long enough to call it prevalent.


So what have I learned, and maybe you all knew it already?

1. A couple hundred Mhz is pretty much negligible for phenom II cores, at least as far as Dolphin is concerned.
2. Even though Dolphin only runs on two cores, excepting LLE on thread. A similarly clocked Phenom II x2 can achieve virtually the same FPS in theses examples however it can't maintain them as well as an x4 or higher.
3. It seems TurboCORE wasn't really doing anything for me. I was using the stock HSF though. Ill test again with better cooling.

I am curious as to how a similarly clocked triple core would fair. If i see any significant gains while OCing this cpu Ill post results. I'm Prime95 testing it right now at 3.4.

(12-21-2012, 05:44 AM)Xalphenos Wrote: [ -> ]2. Even though Dolphin only runs on two cores, excepting LLE on thread. A similarly clocked Phenom II x2 can achieve virtually the same FPS in theses examples however it can't maintain them as well as an x4 or higher.

The lesser frame rate stability may be due to background programs and/or services.

Yeah. I did expect the quad to do better for that very reason. I just didn't expect it to do so well as a dual.
I'm not posting pics but for general information here are the results at 3.4ghz.
1. rock solid 47 FPS
2. 39 FPS with a 2 FPS deviation both ways.
3. mostly 60 FPS but between 55 - 62.

I may try for 3.7

Well, different people have different OS configurations and therefore a different amount of background programs and services running. The difference is even more dramatic if you compare an OS config that's been intentionally slimmed-down to one that hasn't really been taken care of at all.

Dolphin itself does technically use more than 2 threads. Although the extra threads don't need very much processing power. You can sometimes see dolphin accounting for more than 50% of total cpu load on quad core cpus. Plus there are other processes that run services that dolphin depends on whose load may increase when running dolphin. Overall I've seen dolphin gain up to a 20% performance advantage on quad core cpus with the same architecture/cache/clock rates without openMP or LLE on thread.

I never went higher than 3.4 with that 550. My brother was happy there and didn't want me to push it. So I gave him my motherboard and put my main rig together after getting it out of storage. So here are some more benchmarks.

OS: Windows 8 Pro x64
CPU: Core i5 2500k (Stock, occasional turbo to 3.4) EDIT: I was wrong. Turbos to 3.4 on all cores as soon as game starts with occasional spikes to 3.5.
GPU: Gigabyte GTX 660 2GB GDDR5
RAM: 8GB G.Skill 1600 DDR3
Dolphin: 64bit From this thread only options changed where the ones Starscream changed.
GAME: NTSC

OS: W7 Ultimate x64
CPU: 2600k oc'd to 4.6GHz
GPU: GTX 670 [02G-P4-2678-KR]
RAM: Corsair Dominator Platinum 16gb [CMD16GX3M2A1866C9] oc'd to DDR3-2133 with Cas Latency of 9
Dolphin config: x64, no changes

RESULT (sorry for small text! still readable?)


EDIT:
1st screenshot: FPS: 90 - VPS: 180 - SPEED: 300%
2nd screenshot: FPS: 74 - VPS: 148 - SPEED: 246%
3rd screenshot: FPS: 100 - VPS: 201 - SPEED: 335%

(01-12-2013, 04:51 PM)Lucario Wrote: [ -> ]sorry for small text! still readable?

And this is why the Dolphin build in the first post really should have "Auto adjust Window Size" enabled by default. I honestly have got to wonder why it wasn't from the beginning...

(01-12-2013, 11:14 PM)Nintendo Maniac 64 Wrote: [ -> ]

(01-12-2013, 04:51 PM)Lucario Wrote: [ -> ]sorry for small text! still readable?

And this is why the Dolphin build in the first post really should have "Auto adjust Window Size" enabled by default. I honestly have got to wonder why it wasn't from the beginning...

I was referring to the title bar. You could barely read numbers next to FPS, VPS, and SPEED.