Benchmarking Dolphin

[NaturalViolence]

Quote:I may just start a thread named "BENCHMARKING DOLPHIN STARTING RIGHT NOW".

By all means if you want to. Hell if anyone wants to pick this up/take over I'll gladly let them continue in this thread. It will save you time and reduce confusion. Although if someone does decide to do this, please out of respect/kindness

1. TELL ME! That is very important, either send me a private message or email.

2. Keep me in the loop, I want to be informed when we get a "breakthrough".

3. If it's alright I would like to include some instructions on how the benchmark should be cunducted and specific test setups since there are a few things that I personally want to find out from this. For example the effect of hyperthreading must be observed even though we know what the end result will be, it is still important none the less, to do this simple take someone with an i5 750 and an i7 870 and clock them at the same clock rates (that way the only difference is HT).

or you could just be patient and wait for me to get back home . Or you could charge temporarily and give control back to me when I get back. I have said all of this for 2 reasons. 1. I know how impatient some of you are and this must be driving you guys nuts and 2. I know I'm not the only person who wants to do this. Although obviously I would like to be the guy who does this since I already planned everything out except for the game, but what I'm saying is I'm willing to turn it over if you really want and those are my reasons why.

[Starscream]

Just curious if you ever plan on doing this. I may eventually start a new thread if you do not hold your word, but the test will be done differently. If I decide to make another thread, I will not be doing the testing as you were planning to and if you want to be in the "loop" you'll have to get yourself involved.

[NaturalViolence]

Quote:Just curious if you ever plan on doing this. I may eventually start a new thread if you do not hold your word, but the test will be done differently. If I decide to make another thread, I will not be doing the testing as you were planning to and if you want to be in the "loop" you'll have to get yourself involved.

Fair enough but I don't see how I will have the time to do a full blown benchmark this month, I'm in the middle of finals and I have a lot of stuff to deal with after that (moving to a new university). I really can't get this started until early may.

Well that's not a completely fair statement now that I think about it. I could get it started right now if I wanted but that's if I wanted to go the lazy route, which I don't. To get it done right will require more time. I'm just afraid then when someone else gets it started up they won't do it properly and we'll have inconsistent and unusable data as a result (been down this road before) of a failure to account for all variables both software and hardware.

Edit: Ok you know what screw it, I'll start it now. Unfortunately I'm going to have to throw out the results of the vote since mario galaxy is nearly unusable for this benchmark for a variety of reasons. So the runner up, twilight princess wins. Everyone who still wants to participate and has twilight princess please post which platform (GC or Wii) and which region (PAL or NTSC) you have it for.

Edit again: IRC channel is now up.

irc.unvanquished.net
#dolphinbenchmark
Port 6667

[JADS]

Gamecube - NTSC Version
Wii - PAL Version

[doubleffect]

Well, I never played none of these games so there should be one more option. I'm more interested in games that were released for multiple platforms mostly (+ all exclusive Resident Evil games) and if I can't play them on other systems, Dolphin didn't let me down so far and I've got Intel C2D E7400 now @3.48 GHz (well, cache is pretty low now in my opinion, only 3MB). But I really didn't test a lot of games.

[schez]

Why not just use homebrew, source codes are also very often available so u could just nop out all kinds of menus etc. out of the code.
My idea was rather to modify dolphin to get a benchmark button that will be very simple made to code:
It just starts the game in an unthrottled mode, and disables all input to the homebrew, and after 30 seconds it will show u a total score which consists of all drawn fps and additionaly it will show you the an avg fps.
That is very easy to do programming-wise, someone would just have to find a good free homebrew game or just some kind of homebrew demo/intro, that runs on dolphin and is using 3D functionality (it doesnt have to be very complex, higher fps means also more consistent scores to compare with).

[boogerlad]

agreed. I think that if it's homebrew, it should just be a very complex rendering with a lot of cpu and gpu used, along with sound, with the ability to turn several things off to test each one. If using a game, it would be hard to benchmark unless it was a controlled time demo with the same buttons/key input being the same for each person to benchmark.

[NaturalViolence]

Platform: Windows
Dolphin Build: v2.0 64 bit
Game: The Legend of Zelda the Twilight Princess
Version: Any, but you have to tell me or you can't participate
Spot: Faron woods (this may change later, still getting the save game ready)
Settings:
Global:
Enable Dual Core
Enable Idle Skipping
HLE The IPL
JIT Recompiler
Uncheck lcok threads to cores, dsplle on thread, confirm on stop, and panic handlers
Plugins:
Direct3D9
DSP-HLE
GCPAD or Wiimote
Graphics:
Fullscreen
Aspect ratio force 4:3
Resolution: 640 x 480 on fullscreen
Anti-alias: None
Enable CPU EFB Access
Disable Fog
Overlay fps counter
Enable efb copy to ram
Uncheck safe texture cache, everything in the enhancements tab, use XFB, and use real xfb
Audio:
Dsound
Enable HLE Audio
Enable DTK Music
Uncheck enable audio throttle and RE0 audio hack

[Starscream]

I'm afraid I'm not going to participate if it's going to be done with a 32-bit build. I understand that it is easier using a 32-bit so that everyone here can participate, but I prefer to use 64-bit builds.

Good luck with everything.

[hakunushi]

32 bit should be past XD