|Page (1) of 2 - 04/12/06||email article||print page|
MacBook Pro Benchmarks 2: Adobe Apps in RosettaPhotoshop, Illustrator and After Effects lose in 'translation'
Now, if you read our previous installment in the benchmark series, you were no doubt impressed by the results of Intel-native software running on the MacBook Pro. Apple's Final Cut Studio suite performed not only admirably, but it soundly trounced a dual-processor (2.0 GHz) G5 in several tests involving rendering and encoding. (If you have not read that article, you can find it by clicking here.)
Given that the MacBook, running off a mobile Intel processor, can so clearly deliver performance that a typical (if currently low-end) G5 has trouble matching, what do we care about the performance of software that doesn't run natively? The answer is that we're stuck with non-native software for a good, long while yet. And, in the case of Adobe software, that's a critical consideration. Most of us are, after all, utterly dependent upon Adobe for our daily bread; and even those who aren't dependent are reliant on Adobe in one way or another. Given this, and given that Adobe has no known plans for an Intel port of its creative software anytime soon, this is information you need to know if you're considering switching to an Intel-based Mac prior to the release of Universal Binary versions of Adobe's software.
So, to give you an idea of how such a switch might impact your workflow, we've conducted some benchmark tests running After Effects (6.5 and 7.0), Photoshop CS2 and Illustrator CS2 on a 2.16 GHz dual-core MacBook Pro and compared with with other systems.
How'd the MacBook hold up?
Not so well. Whereas in our tests on Intel-native Final Cut Studio software, the MacBook Pro beat a dual 2 GHz G5 desktop in almost every test, in our tests involving PowerPC versions of Adobe's software, the MacBook lost in every single test--often dramatically.
The MacBook Pro used in these tests is equipped with a 2.16 GHz dual-core processor, 2 GB RAM and a 256 MB ATI X1600 graphics card. Other systems tested are comparably equipped desktop systems. Once again I want to emphasize that this is not a review of the MacBook Pro, just a part of our ongoing series of benchmark tests on the system.
In the case of Photoshop CS2, there's good news and bad. The bad news is that Photoshop running under Rosetta's emulation technology performs at about half the speed of a dual 2 GHz G5 desktop system, which, in turn, runs at about half the speed of a top-end G5 Quad desktop. The good news is that, in day to day operation, the effects of this slowdown won't be terribly noticeable in many cases. For example, if a Gaussian Blur filter takes one-sixteenth of a second to render on a G5, it takes one-eighth of a second to render on a MacBook Pro. Certainly nothing you'd notice in your daily workflow. In the case of filters and image adjustments, the slowdown only becomes noticeable when taken in cumulatively, as we'll get to below.
But there is one area where the slowdown is immediately apparent, and that is in transformations--rotating a canvas, rotating a layer, resizing, etc. Here you will see the unfamiliar progress bars indicating that the kind of number crunching that goes into transformations may be a bit much for Rosetta's "translation" to handle efficiently.
Here are the numbers. Explanations follow.
Test 1 involved the creation of a 4,000 x 4,000-pixel document, the application of a few filters and several transformations, including rotating layers and rotating the canvas. This test included 28 individual filters plus 19 image adjustments, transformations and various other functions. That's 47 commands applied in sequence in a single action. On the MacBook Pro, those functions averaged out to slightly more than two seconds each--not terrible considering the size of the image to which they were applied, but certainly not nearly as good as the other machines for which we have benchmarks for these tests.
Test 2 involved the creation of a 2,000 x 1,500-pixel document and more transformations than the first test. This was the most intensive of the tests for the MacBook, with each transformation taking a considerably more time than any other machine tested. There were 30 total commands in this test, with an average time of almost nine seconds per command.
Test 3 included the application of every single filter that ships with Photoshop CS2, with the exception of Displace. It also included a few transformations, text manipulation, selections and fills. The image size for this test was 800 x 600 pixels. There were 123 total commands in this test, averaging about 0.6 seconds per command on the MacBook Pro. This filter-heavy test shows that the MacBook can operate at reasonable speeds in Photoshop, as long as there aren't too many complex transformations involved.
After Effect 6.5 and 7.0
Now, with Photoshop, whatever slowdown you experience from the emulation of Rosetta is, in my opinion, within the realm of tolerability. It's not really much of a step backwards when going from a G5 system when the operations are taken individually--a half second delay here, a two-second delay there. Nothing to make you pull your hair out over. But with After Effects, the delays are obviously compounded frame by frame, effect by effect whenever you render out a project. And here you'll experience some serious lag in Rosetta--and this applies to After Effects 6.5 and 7.0. All things being equal, based on previous tests of software written natively for Intel-based Macs, you'd expect these render times to be cut by more than half if After Effects were running natively.
Here are the bare numbers. Again, keep in mind that we're comparing a machine based on a mobile processor running non-native software against desktop systems running native software.
Test 1 was a simple cel-style animation that involves a PICT file and tracing paths. This is a fairly simple test, and the MacBook didn't lag far behind the dual-processor G5, although it took twice as long as the G5 Quad to render.
Test 2 was a composite using a variety of effects. (Note that the second test could not be rendered in After Effects 7 without modification, so it was left out here.) Again, the lag behind the dual-processor G5 wasn't too substantial--the G5 rendered it in about 70 percent the time it took the MacBook Pro--but in comparison with all the other desktop systems, it was downright sluggish.
Related Keywords:macbook pro, benchmarks, performance, rosetta, after effects, photoshop
Source:Digital Media Online. All Rights Reserved