Dolphin, the GameCube and Wii emulator - Forums

AnyOldName3: You seem to have turned this thread into your own private computer build/advice thread. Please move that conversation somewhere else.

I didn't do it on purpose, honest.

I have tried to move it to random, but I'm still being asked questions here. If anyone wants to ask me something, or give me advice, please do it there.

Trying this out again now that I have all my main pc components together.

I noticed that the fps numbers can have some pretty random spikes at times. How can we be sure that some of the numbers posted by members are actual averages and not some good timing screengrabs? For example, I usually hover at around 97-98fps for save 1 (4.6ghz 3570k) but I could also catch a frame where I'm at 100fps if I wait for the right time, monitoring the framerate.

Unfortunately there's no guarantee. A lot of members have also posted results where the boat is visible.
Best IMO would be to use the built-in benchmark tool (or FRAPS/any tool that has this feature) of Dolphin for these type of benchmarks. (We'll have to move to 3.5+ though)

Guess it'll be time to update the sheet.

Edit: There we go. Updated. A lot has been off-topic talk .

georaldc Wrote:How can we be sure that some of the numbers posted by members are actual averages and not some good timing screengrabs? For example, I usually hover at around 97-98fps for save 1 (4.6ghz 3570k) but I could also catch a frame where I'm at 100fps if I wait for the right time, monitoring the framerate.

There is always going to be some small variation. But any large variation can be weeded out by comparing against other similar cpus. We've already invalidated a few results by doing this.

Wait, we've been finding averages? I just pushed PrtScn as close as possible second after loading each save. Do I need to now fix my results by getting an average over say 30 seconds, or will they still do?

I've always tried to get a good average when I did benchmarks with this thread, but When doing personal benchmarks for myself to test my different computers and settings, I use a new build, use FPS logging, and have the benchmark run for a set time before stopping it and taking the average of the FPS. Otherwise, testing stuff like whether ICC builds make any difference in performance would be too difficult. And if you're curious about the results of that, an ICC build with all Haswell optimizations enabled only improved performance by 2-3% over a standard Visual Studio build.

Well at least you've finally settled the 'Do they actually make things slower' debate.

Maybe it's a bit late to participate, but here are my results, anyway...

HP Envy dv6t-7300 Laptop
CPU: Intel Core i7 3630QM @ 2.4GHz
GPU: nVidia GeForce GT 635M @ 2 GB GDDR3
RAM: 8 GB @ DDR3-1600
OS: Windows 8 (RTM) [x64]

While running these tests, the native resolution was 1920x1080 and according to CPU-Z, the processor speed was at 3.2GHz

Quest log 1: 64FPS, 128VPS and speed at 213%


Quest log 2: 52FPS, 103VPS and speed at 171%


Quest log 3: 79FPS, 159VPS and speed at 265%

(07-03-2013, 07:37 AM)AnyOldName3 Wrote: [ -> ]Wait, we've been finding averages? I just pushed PrtScn as close as possible second after loading each save.

You have to at least let the fps stabilize from the initial game load, then take the shot. It's okay if you use the higher fps number, but it shouldn't be more than a 2-5fps difference as the fps jumps around.

(07-06-2013, 02:38 AM)Jhonn Wrote: [ -> ]Maybe it's a bit late to participate, but here are my results, anyway...

It's never too late.