|Page (1) of 1 - 04/21/06||email article||print page|
Benchmarks: Windows XP on the MacBook Pro, Part 2LightWave, Maya and Cinema 4D performance tests
For this latest course of benchmarks, we return to three applications we tested previously on the MacBook running Mac OS X: Maxon Cinema 4D, NewTek LightWave and Autodesk/Alias Maya. On Mac OS X, Maya and LightWave were running in emulation under Apple's Rosetta technology. Being non-native applications, they took quite a performance hit in terms of rendering scenes. But under Windows XP, they're running natively, so we get to see the full processing potential of the Intel CoreDuo that powers the MacBook.
Cinema 4D, of course, if fully native on PowerPC and Intel sstems running both Mac OS X and Windows. However, since our first benchmark test, a new version of Cinebench has been made available to us (Cinebench 9.6), producing dramatically improved results running under Mac OS X on Intel hardware. The updated benchmark figures are included here, as well as new benchmarks for Cinebench running under Windows XP.
Before we get started, if you'd care to go back and take a look at our previous benchmark analyses, you can do so by following the links below.
- MacBook Pro Benchmarks: Final Cut Studio 5.1
- MacBook Pro Benchmarks 2: Adobe Apps in Rosetta
- MacBook Pro Benchmarks 3: 3D Performance
- Benchmarks: Windows XP on the MacBook Pro
Now, on with the new tests.
Maxon Cinema 4D
In our previous tests of Maxon's software, we ran Cinebench 9.5, which was the latest version of Maxon's benchmarking utility available at the time. However, since then, Maxon has updated Cinema 4D to version 9.6 and has allowed us to use a new 9.6 release of Cinebench as well, which, as of this writing, isn't yet available publicly. But it does reflect changes in Cinema 4D R9.6 that produce dramatically improved results in Mac OS X, particularly in OpenGL hardware acceleration. As these tests show, where once there was a disparity in performance in favor of Windows, there's now one in favor of Mac OS X running on the same hardware.
All results for these tests are shown in units defined by Cinebench for cross-platform reference. Higher numbers are better. Most of the results here are derived from Cinebench 9.5 or earlier, and only the MacBook was tested with Cinebench 9.6. . The latest results give a boost to the MacBook running OS X. In the previous test, the MacBook received a CPU score of 611. Now it's at 627. And the graphics test showed a result of 1431 CB-GFX in OpenGL hardware. Now that's been boosted almost incredibly to 2764. OPenGL software and Cinema 4D shading also saw dramatic increases.
In all the tests, the Mac S X version of Cinebench outperformed the Windows version, although that could easily change with any new release of the software. It does show that, at least for now, Cinema 4D will perform better when running under OS X than Windows XP on the exact same hardware.
As far as the comparison with other machines goes, there are no real surprises here. The MacBook is outperformed in terms of CPU score by all of the quad-processor systems, as well as by the dual-processor 3.6 GHz Xeon system, though it wasn't far from matching that one in the latest round of tests. It is interesting to note, once again, that on a per-processing-core basis, the MacBook did seem to beat out everything but the Xeon.
Not surprisingly, the MacBook Pro produced much better results in our LightWave tests running under Windows XP than it did running under Mac OS X. Unlike Cinema 4D, LightWave still doesn't run natively under Mac OS X on Intel hardware. Instead, it runs in emulation under Apple's Rosetta. The performance hit for that is quite extreme, costing up to a 75 percent loss in performing power. The results on the MacBook running Windows give a more accurate view of the true potential of this hardware and how LightWave will perform once a native version becomes available for Intel-based Macs. The Windows version beat the dual 2.0 GHz G5 in every single test (though in some cases only marginally), while it was beaten in every test by the 2.5 GHz G5 Quad.
All of these tests are based on rendering times on scenes that are included with LightWave.
Test 1 rendered out the scene called Radiosity_BOX.lws (/Content/Scenes/Benchmark/). It contains 7,294 polygons and renders at 600 x 600 with anti-aliasing set to "Low" and radiosity turned on. This took about 23.5 minutes rendering in WIndows and more than an hour and a quarter running in Rosetta under Mac OS X.
The second scene is SunsetSample.lws (/Content/Scenes/). This is the simplest of the scenes tested and includes one polygon with four points. It's rendered with no anti-aliasing and no radiosity. The Windows version of LightWave rendered this scene in a quarter of the time it took the OS X version.
The third test involved the scene called Teapot.lws (/Content/Scenes/Benchmark/). It contains 25,938 polygons and renders with anti-aliasing set to "medium." Radiosity is off for this test. It took the Windows version of LightWave about a third the time it took the version running under Rosetta.
The fourth test uses The_Matrix5.lws (/Content/Scenes/Abstract/). This is the most processor-intensive of all the tests. It involves 24,288 polygons and renders with anti-aliasing set to "enhanced low" and radiosity off. The render resolution is 800 x 400. It took the MacBook Pro a little more than 4 hours, 45 minutes to render a single frame of this scene under Rosetta, while it took the Windows version a scant two hours. The G5 Quad did it in a ridiculous 43 minutes.
The final test involved depth of field. It uses the Virus_DOF.lws scene (/content/Scenes/Surface/). Again, no contest between the Windows version and the Rosetta version.
And, finally, we come to our Maya tests. Again, no surprises here, at least not judging from our previous tests. The MacBook Pro beat the dual 2.0 GHz G5 in every single test, and it soundly defeated the Maya running under Rosetta on the same hardware. Maya, like LightWave, is not yet available in a version that runs natively under Mac OS X on Intel hardware, at least as of this writing.
Test 1 involved rendering a smoke-filled scene at 320 x 240 with highest-quality anti-aliasing. Under Windows, Maya rendered it in less than one-third the time it took under Rosetta.
Test 2 rendered a fog scene at 200 x 200 with anti-aliasing set to high quality. This time, the Windows version handled the job in a quarter the time it took the Rosetta version. It also handled it in less than half the time it took the dual 2.0 GHz G5.
Test 3 used a scene with a soft-body object and particles. It was rendered at 640 x 480 with anti-aliasing set to high quality. In this test, the Windows version once again handled the task in better than one-third the time it took the Rosetta version.
Test 4 involved rendering a cutaway scene of a building that included a number of several objects. It was rendered at 768 x 415 with anti-aliasing set to "highest quality." In this test, the MacBook barely outperformed the G5, though, again, it was more than twice as fast in Windows than in Rosetta under Mac OS X.
And the final test involved rendering a complex object at 720 x 486 with anti-aliasing again set to "highest quality." This was another test that showed the MacBook just barely faster than the G5. The Windows version handled it in less than half the time it took under the Mac OS X version.
Our final benchmark analysis confirms about the MacBook Pro what we've been seeing from the beginning. In tests where the software is running natively, this hardware proves itself a match for desktop systems. But in the case of 3D software, the results aren't as dramatic as they were in our tests of After Effects, Photoshop and Illustrator. In those tests, the MacBook matched or even exceeded the performance of quad-processor workstations. Here, it surpassed the dual 2.0 GHz G5 in every test, but it couldn't match the G5 Quad or other quad-processor systems in terms of CPU performance. We will, of course, have to wait for Intel-native Mac OS X versions of Maya and LightWave to see if there's any performance gain. Until then, it's clear that the MacBook is capable of handling every hard-core test that can be thrown at it from professional-level creative applications. It's better than some desktops, and it can even compete with top-end workstations in some cases.
Coming up next, we'll have a formal review of the 2.16 GHz MacBook Pro. We'll also provide updates on benchmarks as Intel-native versions of After Effects, Photoshop, Illustrator, LightWave and Maya become available. In the meantime, if you have any further questions, be sure to drop me a line or visit me in one of the forums listed below.
Related Keywords:macbook pro, benchmarks, 3d, windows
Source:Digital Media Online. All Rights Reserved