New Dolphin CPU Benchmark - NO GAME REQUIRED

[Zee530]

(03-12-2014, 07:54 PM)Lucario Wrote:

(03-12-2014, 01:36 PM)NaturalViolence Wrote: How the hell would anyone know that?

1. use same CPU & clockspeed for this test
2. find average lowest FPS in Hyrule field (where? savestate possible?)
3. find time in benchmark
4. convert minutes to seconds for easy multiplication
5. 30/FPS = FPS2
6. time/FPS2 = time2
7. convert time2 to minute

Repeat with different CPU and/or clockspeed to see if it's consistent? If so, we've prolly found ideal time that will run hyrule field at full speed.

Doesn't this assume that the GPU does absolutely no work whatsoever?

[Lucario]

Right, and it shouldn't be supposed to bottleneck during Hyrule Field.

I think it's alright to use latest Dolphin build for Hyrule Field, while using old build for time benchmark.

[NaturalViolence]

Lucario Wrote:

NaturalViolence Wrote:How the hell would anyone know that?

1. use same CPU & clockspeed for this test
2. find average lowest FPS in Hyrule field (where? savestate possible?)
3. find time in benchmark
4. convert minutes to seconds for easy multiplication
5. 30/FPS = FPS2
6. time/FPS2 = time2
7. convert time2 to minute

Repeat with different CPU and/or clockspeed to see if it's consistent? If so, we've prolly found ideal time that will run hyrule field at full speed.

Ok let me rephrase: "How the hell would anyone here already know that?"

You asked as if you're sure somebody here already knows. Which is a completely unreasonable assumption.

And while your method is reasonable it's not going to yield perfectly accurate results. Differences in OS, settings, build, NTSC vs. PAL, and GC vs. Wii version will all effect hyrule fields performance and can't be accounted for with this benchmark. Then there is the issue of scaling and workload differences. A cpu may be 50% faster than another cpu in this benchmark but only 10% faster in hyrule field for example. The results in these two different workloads may not and probably will not line up. And then finally there is the issue of the GPU as Zee rightfully pointed out. Just because it's not the biggest bottleneck doesn't mean it has no impact on performance at all (particularly when you factor in different drivers, OS, graphics setting, and backend). You'll probably get "pretty good" results that give you a general idea with your method but certainly not a solid universal number that you could throw around to immediately tell how someones hardware would perform.

[Lucario]

Oh alright, we can start test that now, and like I said, to see if test results are consistent between different CPUs.

If the ideal time much varies between different CPUs (50% for time benchmark and 10% for hyrule field just like you said) then you're right, it doesn't work out.

I'm doing test right now and the vague ideal time seemed to be 5:39. I'll run test on different clockspeed and use different GPU (much weaker).

EDIT: Apparently HD 5570 couldn't handle Hyrule Field very well like my GTX 670. Interesting. Does Hyrule Field FPS still varies between high end GPUs as well?

CPU: i7 2600k
GPU: GTX 670
Dolphin build: 4.0-1146 running gamecube version (GZ2E01)

clockspeed: 4.5ghz
Benchmark time: 598 seconds
Hyrule Field highest FPS: 20
Hyrule Field lowest FPS: 17
Ideal time: 5:39

clockspeed: 4.0ghz
Benchmark time: 664 seconds
Hyrule Field highest FPS: 18
Hyrule Field lowest FPS: 16
Ideal time: 5:54

clockspeed: 3.8ghz
Benchmark time: 703 seconds
Hyrule Field highest FPS: 17
Hyrule Field lowest FPS: 15
Ideal time: 5:51

[ExtremeDude2]

CPU: AMD Phenom II x4 955 BE @ 3.2GHz
OS: Windows 7 Home 64-bit
Time: 18 minutes, 19 seconds
Screenshot: http://i.imgur.com/XvleZyV.png

[tuanming]

(03-13-2014, 04:09 PM)Lucario Wrote: Oh alright, we can start test that now, and like I said, to see if test results are consistent between different CPUs.

If the ideal time much varies between different CPUs (50% for time benchmark and 10% for hyrule field just like you said) then you're right, it doesn't work out.

I'm doing test right now and the vague ideal time seemed to be 5:39. I'll run test on different clockspeed and use different GPU (much weaker).

EDIT: Apparently HD 5570 couldn't handle Hyrule Field very well like my GTX 670. Interesting. Does Hyrule Field FPS still varies between high end GPUs as well?

CPU: i7 2600k
GPU: GTX 670
Dolphin build: 4.0-1146 running gamecube version (GZ2E01)

clockspeed: 4.5ghz
Benchmark time: 598 seconds
Hyrule Field highest FPS: 20
Hyrule Field lowest FPS: 17
Ideal time: 5:39

clockspeed: 4.0ghz
Benchmark time: 664 seconds
Hyrule Field highest FPS: 18
Hyrule Field lowest FPS: 16
Ideal time: 5:54

clockspeed: 3.8ghz
Benchmark time: 703 seconds
Hyrule Field highest FPS: 17
Hyrule Field lowest FPS: 15
Ideal time: 5:51

I'll bench with you and you can make a new thread. Also post your savestate and let's use the latest official build. However, I like to do this in a more simple method. List the lowest, average, and high fps. I can downclock my speed to match your 4.5GHz 2600k.

Specs:
CPU: 4770K OC'ed at 4.8GHz
GPU: N580 Lightning Extreme 3GB,
RAM: CORSAIR Vengeance Pro 8GB DDR3 2666MHz
HDD: Samsung 840 Pro SSD 512GB
CPU cooler: Noctua NH-U14S w/Noctua NF-A15 PWM 140mm Premium
MOBO: Asus Maximus VI Hero LGA 1150 Z87

[Lucario]

(03-14-2014, 09:24 AM)tuanming Wrote: I'll bench with you and you can make a new thread. Also post your savestate and let's use the latest official build. However, I like to do this in a more simple method. List the lowest, average, and high fps. I can downclock my speed to match your 4.5GHz 2600k.

Specs:
CPU: 4770K OC'ed at 4.8GHz
GPU: N580 Lightning Extreme 3GB,
RAM: CORSAIR Vengeance Pro 8GB DDR3 2666MHz
HDD: Samsung 840 Pro SSD 512GB
CPU cooler: Noctua NH-U14S w/Noctua NF-A15 PWM 140mm Premium
MOBO: Asus Maximus VI Hero LGA 1150 Z87

Just realized DX11 was at fault for slowing Hyrule Field down alot from OpenGL

Does this backend has anything to do with CPU side but not to GPU? I just saw different FPS corresponding to my 2600k's different clock speeds.

Don't think I'd make new thread for this...

If you're still interested in doing benchmark and compare against my 2600k, here it is.

REQ SETUP
Dolphin version: 4.0-1146
Backend: DX11
Save state: http://www.mediafire.com/download/lzb1t4b188n037d/GZ2E01.s01
Game: LoZ Twilight Princess [GZ2E01]

[delroth]

Move your bad benchmark to another thread please. I'll remove any following posts about it.

[Stead]

CPU: Intel Xeon E3 1230 V2 @ 3.3 + Turbo boost 3.5
OS: Windows 8.1 Pro x64 with hyper v enabled
Time: 11 Minutes 45 Seconds.
Screenshot: http://i.imgur.com/A1K2DQI.png

[AnyOldName3]

That is interesting. Link to the actual article?