|Page (1) of 1 - 07/26/06||email article||print page|
Shake 4.1 Core Duo BenchmarksTests of Shake's rendering performance on Apple Intel hardware
In our recent review of Apple's Shake (which you can read by clicking here), I mentioned that I tested the software on a 13-inch MacBook (2.0 GHz white model) and found the performance to be more than adequate.
As we've seen in previous benchmarks on the 13-inch MacBook, CPU-intensive tasks like rendering are handled extraordinarily well by the MacBook's 2.0 GHz Intel Core Duo processor. But, in the case of Shake, the results are extraordinary.
With some other applications, we saw the MacBook keeping up with G5 workstations and often outperforming them. But the margins here are just incredible, demonstrating that Shake on a notebook system is not only viable, but clearly a better way to go than a lower-end dual-processor G5.
Before we get to the numbers, if you would like to view any of our previous benchmarks of the 13-inch MacBook, you can find them by clicking on the links below.
- Benchmarks 1: Final Cut Studio
- Benchmarks 2: Adobe Apps in Rosetta
- Benchmarks 3: LightWave, Maya and Cinema 4D (Rosetta and Native)
- Benchmarks 4: Adobe Photoshop, Illustrator and After Effects in Windows XP
- Benchmarks 5: Cinema 4D, LightWave and Maya in Windows XP
And now for the numbers. With Shake 4.1, I went ahead and ran benchmarks on two systems (the two systems I currently have available to me). Those were a dual 2.0 GHz G5 with 2 GB RAM and a 13-inch MacBook 2.0 GHz with 2 GB RAM. I ran four tests on these two systems in Shake 4.1 and also ran the same tests in Shake 4.0 on the G5. (Shake 4.0 was not a Universal Binary, so it couldn't be tested on the MacBook. Please no letters about Shake not running on a MacBook. I received enough of those about Final Cut Studio. As with Final Cut Studio, the MacBook will only run the Universal Binary version.)
These tests measure only the rendering performance of the G5 and MacBook, since this is easily quantifiable and provides a good measure of CPU performance--which is important not just for the primary Shake application, but also for situations where you might be looking to add machines to a render farm. OpenGL performance is obviously going to be superior on a system that has a hardware-accelerated OpenGL card, which the 13-inch MacBook does not. (But the MacBook Pro does, and, as I've mentioned in past articles, the 15- and 17-inch MacBook Pro models can do real-time 2k compositing in Shake.)
Here are the CPU numbers. Explanations follow.
All of these tests record the amount of time it took Shake to render the individual projects using the "Render Disk FlipBook" command. In every case, the MacBook beat out the G5 drastically.
For test 1, I used a single source video file, a piece of HD footage (720p). To this I applied the following nodes from the "Filter" category: PercentBlur, |DilateErode, ApplyFilter (two instances) and Grain (two instances). SHake 4.1 actiually ran a bit slower than 4.0 on the G5. Shake 4.1 on the MacBook drastically outperformed both versions on the G5.
Test 2 involved a piece of footage in the DVCPro HD 720p60 format. This test used various nodes in the "Color" category, including AdjustHSV, Compress, Solarize and Set. Again, Shake 4.1 was slower than 4.0 on the G5, but drastically faster on the 13-inch MacBook. (Remember, this is a 13-inch MacBook, not a MacBook Pro, so the optimizations in Shake for the Intel hardware are phenomenal.) The MacBook completed the render in 71 percent of the G5's faster time.
For test 3, I switched to some D1 footage and applied three nodes from the "Transform" category: CameraShake, Move2D and Zoom. The MacBook rendered the project in exactly half the time it took the G5 to render the same project.
And for test 4, I used a piece of uncompressed HD (720p) footage with a |Displace node appied to it. Displacing the footage is a RotoShape with the following nodes applied to it: AddShadow, LensWarp, PercentBlur and AdjustHSV. This was the only project in which Shake 4.1 beat Shake 4.0 on the G5. The MacBook, once again, came out on top, this time rendering the project in about 69 percent the time it took Shake 4.1 on the G5 and half the time it took Shake 4.0 on the G5.
(It's worth mentioning, in case you didn't realize it already, that a dual 2.0 GHz G5 is not the fastest PowerPC-based machine on the market. It's simply the G5 that I use as a reference machine. In terms of CPU performance, it's easy enough to extrapolate the numbers for other models, since, in my experience, performance is fairly consistent in the G5 line in terms of the speed differences in relationship to the total GHz of any given system. For example, in many of my CPU tests in the past, a G5 Quad, with 10 GHz total, proved to be about 2.5 times the speed of my dual 2.0 GHz G5 (with its 4.0 GHz total), although there can be some variation based on the software involved and the operations being performed. If you wish, you can read an article comparing CPU performance of the G5 Quad with the G5 used for these tests by clicking here.)
So what do the numbers mean? If you're sticking with a PowerPC-based machine for now, an upgrade to Shake 4.1 isn't all that vital. Some functions will be faster; some will be slower. And, beyond the Universal Binary port, there's nothing new about Shake to justify the upgrade, though, as I mentioned in my review, this is by far the lowest price for a Shake upgrade in its history.
But, if you work with Shake and are planning to migrate to an Intel-bsed system, or, if you have an Intel-based system and are thinking of acquiring a Shake license, there couldn't be a better time. Shake is now at its lowest price ever, $499 for the full version. And Shake 4.1 on an Intel-based Mac is a speed demon. I can't even imagine what the numbers will be like when Apple comes out with its Intel-based workstations.
Related Keywords:shake 4.1, apple shake, review, benchmarks, intel, macbook, macbook pro, g5, comparison, performance, Visual artists, film professionals, effects houses, studios
Source:Digital Media Online. All Rights Reserved