Cinebench 9.5 comparisions between OS X and XP on 2.16gHz MacBook Pro

[jann]

[edited to fix typo in labeling]

Just was wondering what the differences were, so i loaded cinebench 9.5 on both the XP side and the OS X side and ran them. Standard VGA driver from XOM.efi and ALL video acceleration on XP turned off (cos the XOM driver does not use the GPU).

The results are interesting:

XP: 1 CPU: 314
OS X: 1 CPU: 328

XP: 2 CPU: 566
OS X: 2 CPU: 604

XP: CPU RATIO: 1.8
OS X: CPU RATIO: 1.84

XP: C4D Shading: 315
OS X: C4D Shading: 374

XP: OpenGL Software-L: 217
OS X: OpenGL Software-L: 832

XP: OpenGL Hardware-L: 144
OS X: OpenGL Hardware-L: 1425

XP: SWL RATIO: .69
OS X: SWL RATIO: 2.24

XP: HWL RATIO: .69
OS X: HWL RATIO: 3.83

(working on other benchmarks tonite)

-Jann

[marjetica]

I did a bit of different and much less extensive testing on a MacMini (Core Solo, 1,5 GHz, 1,25 GB RAM). I took two QuickTime movies (mp4 format), one was 6 MB and the other 16 MB. I exported them to iPod format (m4v) in QuickTime Pro 7.0.4. Conversion time was 1 min 2 s for short file and 3 min 10 s for long file running OS X. Respective timings were 0 min 54 s and 2 min 54 s when running Windows XP.

By the way, in same test iMac G5 performed consistently better than iMac Core Duo. There is huge, consistent and reproducible difference in results when working with QT 7.0.3 and 7.0.4 on all platforms (later version yields better scores).

[thrunner]

http://i28.photobucket.com/albums/c231/thrunner/cinebenchP5LD2.jpg

[jann]

Thrunner,

Thanks, but the reason i started the thread was to compare the MacBook Pro running OS X w/the MacBook Pro running XP to see if XP used the hardware anywhere near as well as OS X does.

Am posting other tests this AM.

Jann

[sandau]

would you mind making a summary for those of us not familiar with the results? What does it all mean...

thanks!

[jann]

would you mind making a summary for those of us not familiar with the results? What does it all mean...

thanks!


Sure.

Here it is:

XP: 1 CPU: 314
OS X: 1 CPU: 328

This is Cinebench's way of saying for the entire rendering process, we are going to use 1 CPU. Graphics chip (GPU) has NO effect on this score.


XP: 2 CPU: 566
OS X: 2 CPU: 604

This is Cinebench's way of saying for the entire rendering process, we are going to use 2 CPU's (or more on a multi-cpu system). On the MBP, this means both cores of 1 CPU. Graphics chip (GPU) has NO effect on this score.


What the above tests state (we are not worried about the actual numbers, just how they relate to each other) is 2 things:

1. Cinebench did a great job of using each OS's dual core functionality to get the most they could. If, for instance, OS X's score was appreciably higher than XP's, it would mean that XP was lagging under this hardware..

2. If XP had an appreciably lower score, It COULD point to the fact that the EFI layer (this is NOT the case) that Blanka and Narf put in place was interfering with the OS (burning CPU cycles)--AGAIN: This is NOT the case.

XP: CPU RATIO: 1.8
OS X: CPU RATIO: 1.84

This is simply how much the computations were speeded up by having a 2nd core (or CPU)

XP: C4D Shading: 315
OS X: C4D Shading: 374

Here is where we get iffy. This test does not use the GPU

XP: OpenGL Software-L: 217
OS X: OpenGL Software-L: 832

XP: OpenGL Hardware-L: 144
OS X: OpenGL Hardware-L: 1425

These are Hardware and Software lighting tests. This translates to "Where does light reflect off the suface of this building when we draw all 4 walls and put shine a spotlight on it." Also, "What areas of the building get lit up and stay lit if we move the building in a circle". OpenGL is MADE for this sort of speedup in calculation.

These tests DO USE the GPU. What, effectively this is saying is that because there is NO OpenGL compatible drivers in the XP system, the CPU is doing all the work...and in the case of the Software Test, this matters ALOT! The OpenGL drivers and subsystem are not installed at all when there is no compatible hardware, thus the speeds on XP are EXTREMELY slow.

The Hardware tests reflect that OS X has a GPU with active OpenGL drivers...and thus BLOW AWAY the XP system. This WILL CHANGE when we get a ATI driver that has OpenGL compatibility.

The reason the Software speedup is so much greater than the hardware speedup is that Cinebench is able to use the 2nd core (to a point) in the Software test.

XP: SWL RATIO: .69
OS X: SWL RATIO: 2.24

XP: HWL RATIO: .69
OS X: HWL RATIO: 3.83

This is simple here. A value of greater than 1 indicates OpenGL speedups have taken place. Less than 1 means that the tests were run NOT using OpenGL. Again, this will change once we have ATI drivers with OpenGL compatibility.


Better?