Dolphin, the GameCube and Wii emulator - Forums

****************************
  The Unofficial Dolphin Benchmark
****************************


Features:
--------------
* Cutting-Edge: Based on a recent Development Build (4.0-5064)
* Standardized: Preconfigured with the most optimal *default* settings.
* Easy to Use: Includes a batch file for easy benchmarking. Just click and run.
* Optimized: Disabled all logging.
* Lightweight: Removed all unnecessary files.
* Idiot-Proof: Made most files read-only.


How to use:
-----------------
1. Download Dolphin-Benchmark-v2-Part1.7z and extract the "Dolphin Benchmark" folder (with all of its contents) to your HDD / SSD.
2. Download Dolphin-Benchmark-v2-Part2.7z and extract the 'Dolphin.exe' file to the "Dolphin Benchmark" folder.
3. Right-click on "bench.bat" to create a shortcut on your Desktop.
4. Double-click on the desktop shortcut to run the benchmark.
5. Wait for the benchmark to finish (3 ~ 15 min., depending on your CPU).
6. Take a screenshot.
7. Post your results here.

NOTE: This "package" is for Windows only, but it is possible to run this benchmark on Linux / OSX as well. See the Official Dolphin Benchmark thread for more info.


Benchmarking Tips:
-------------------------
1. Before running the benchmark, it's recommended to reboot your PC and leave it at the Desktop for 20 ~ 40min.
2. Close all running apps, web browsers and temporarily disable any antimalware suites.
3. Benchmark times are variable, so you may need to do multiple benchmark runs.


Scoring:
------------
"A" - Total time of 4:30 or less. Your CPU is fast enough to run the most demanding titles at full speed *, including double-speed mods with WS AR patches and virtual CPU overclock.
"B" - Total time of 6:00 or less. Your CPU should be enough for full speed * in the more demanding titles, with a few exceptions.
"C" - Total time of 8:30 or less. Your CPU is good for (full speed * in) all the "medium" titles.
"D" - Total time of 11:00 or less. Your CPU is good only for the "light" stuff.

* "Full Speed" in this benchmark means full speed all the time (without dips, even in the most demanding parts), not "full speed most of the time" or "full speed half of the time".

Results:
------------

"A"
Core i7-4790K @ 5.0 GHz = 3:47
Core i7-4790K @ 4.9 GHz = 3:54
Core i7-4790K @ 4.7 GHz = 4:03
Core i5-4670K @ 4.4 GHz = 4:16
Pentium G3258 @ 4.6 GHz = 4:25

"B"
Pentium G3258 @ 4.5 GHz = 4:35
Core i7-4790K @ 4.4 GHz = 4:49
Pentium G3258 @ 4.2 GHz = 5:02
Core i5-4690K @ 3.5 GHz??? = 5:21
Xeon E3-1231 v3 @ 3.4 GHz = 5:33
Core i5-4670K @ 3.8 GHz = 5:36

"C"
Pentium G3258 @ 3.2 GHz = 6:13
Core i5-4440 @ 3.1 GHz = 6:25
Pentium G3220 @ 3.0 GHz = 6:38
Core i7-3770 @ 3.4 GHz = 7:02
Core i5-2500 @ 3.3 GHz = 7:17
Core i7-960 @ 4.0 GHz = 7:18
Core i7-4700MQ @ 3.4 GHz = 7:32
Core i3-4330T @ 3.0 GHz = 7:34
Core i5-750 @ 3.7 GHz = 7:45
Athlon II X2 280 @ 4.1 GHz = 8:25

"D"
Core i7-960 @ 3.2 GHz = 9:09
Phenom II X4 955 @ 3.5 GHz = 9:27
Xeon E5440 @ 3.6 GHz = 9:43
Xeon E5-1603 @ 2.8 GHz = 10:23


NOTE: A Quad-Core CPU performs an additional 10%~15% faster in Dolphin (over a Dual-Core CPU with the same score) due to its ability to offload Dolphin's audio thread and OS background tasks to the additional cores.


Download:
---------------

I ran your new benchmark here is a screen cap of my results.

Kids and their crazy overclocks, these days...

(01-13-2015, 07:00 AM)IceStrike256 Wrote: [ -> ]I ran your new benchmark here is a screen cap of my results.

Wow, is your 4790k stable under prime 95 at that clock because mine is not.

Yeah got a new water cooler and needed to use it to it's fullest potential.

@kirby

I notice you updated the build. You do realize that if you change the build that automatically invalidates any results taken before then since they can no longer be accurately compared with those taken after the change, right?

(01-13-2015, 11:25 AM)NaturalViolence Wrote: [ -> ]I notice you updated the build. You do realize that if you change the build that automatically invalidates any results taken before then since they can no longer be accurately compared with those taken after the change, right?

Reverted to 4.0-5064 and updated the batch file. All previous results are valid now.

Restored the [UNOFFICIAL] marking, and added it to the thread itself. This is absolutely unofficial. An official benchmark is planned for 5.0 for proper testing and analysis.

Is there already a set decision that 5.0 will be the next stable release?

Also removed the "v2.0" thing, since it suggested it's a direct successor to the original benchmark.