Power-hungry computer pros are always looking for tools to speed up time-consuming tasks. Many hotshots look straight to the top of the line when shopping for a new Mac. Apple’s professional tower, the highly configurable Mac Pro, recently added another choice that features eight processing cores, all running at 3GHz. For software programmers eager to help their applications take advantage of multicore systems, or for Mac users who regularly spend hours compressing video or rendering 3-D scenes, purchasing the eight-core Mac Pro makes sense. Our testing, however, shows that most people would be served just as well by a less-expensive quad-core Mac Pro.
While it’s true that processor speeds continue to make modest gains, these days most manufacturers are touting multiple processors on their chips, a change from the past, when processor vendors enjoyed frequent, remarkable leaps in individual processor speeds. In fact, every Mac in Apple’s current product line is using at least dual-core Intel processors–the rest of the Mac Pro processor options feature two dual-core processors running at 2GHz, 2.66GHz, or 3GHz. The new eight-core Mac Pro uses two 3GHz quad-core Intel Xeon processors. The problem with this multicore strategy is that throwing more processors at a job doesn’t always mean that the task will finish faster.
Aside from the new processor option, the Mac Pro remains unchanged, with four available eSATA hard drive bays; two optical drive bays; 1GB of fully buffered, 667MHz ECC DDR memory (expandable to up to 16GB); four PCI Express card slots; and a choice of three different graphics cards. Apple claims the number of possible Mac Pro configurations to be over 33 million. I’m not inclined to check the math, so let’s just take the company’s word for it.
In order to try and isolate the performance difference the new processor makes in the Mac Pro, we used Apple’s recommended configuration for the most part, but swapped the standard two 2.66GHz dual-core processors (the quad-core option) for two 3GHz quad-core processors (the eight-core option)–a US$1,498 upgrade. Our eight-core Mac Pro has 1GB of RAM, a 250GB eSATA hard drive, 16X dual layer SuperDrive, and an Nvidia GeForce 7300GT graphics card with 256MB of dedicated video memory.
By using the same configuration on both machines, it is easy to see where and how the different processors impact performance. In the bulk of our testing, in which tasks were performed one at a time, doubling the processors didn’t make much of a difference. In fact, the eight-core processor was just 4 percent faster than the quad-core 3GHz Mac Pro in our Speedmark overall system benchmark. We use Speedmark to test everything from Mac minis to Mac Pros. Some of Speedmark’s application tests see no benefit from multiple processors, and some, like our Microsoft Office tests, still rely on Apple’s Rosetta translation software to run on the Mac Pro’s Intel processors.
One test that did see great improvement with eight processing cores was our 3-D rendering test using Cinema 4D. Running this test, it’s easy to see how well this program takes advantage of all of the processors in a system. The rendering window splits itself into eight horizontal chunks, and you can watch each area being drawn simultaneously. In this test, the eight-core 3GHz system was 44 percent faster than the quad-core 3GHz model, finishing the job in just 14 seconds; the quad took 25 seconds.
The other test that showed measurable improvement, however modest, was encoding a movie using Compressor 3. Running Compressor 3 with multiple cores gets complicated. Many of the compression codecs are not optimized for multiprocessing, so to get the best performance, you need to use Apple’s Qmaster, a utility included with Final Cut Studio that allows you to set up and share a cluster comprised of your system’s internal processor cores. In the Qmaster control panel, you can set a cluster to handle a specific number of cores by configuring Instances. (You can also use Qmaster to put idle computers on your network to work on tasks, but for our tests we used only our internal processors.) Our DVD encoding test worked best when splitting the processors up into four instances. When we did this, the eight-core machine managed to squeeze out speed improvements of between 9 to 15 percent over the quad-core model.
To see how additional RAM might affect performance, we configured each system with 4GB of RAM and ran the professional application tests again. With the additional memory, the eight-core model ended up winning each of our professional application tests. Adding the RAM helped each computer snip a second or two off of the Photoshop and Compressor tests, but our Cinema 4D render test is almost entirely processor driven and was unaffected by the additional RAM.
To see how multiple cores might help a system run multiple tasks simultaneously, we created something of a torture test: We ran three professional program tests while timing the Photoshop suite. When run alone, there was very little difference between the time it took the eight-core and the quad-core Mac Pro to finish the Photoshop suite. While running this special torture test, the eight-core model took 86 percent longer to complete the Photoshop part of the test. Overall, that model was still 24 percent faster than the quad-core Mac Pro when running the entire multiple-program test with 4GB of RAM. The eight-core Mac Pro was 44 percent faster than our 2.5GHz quad-core G5 Power Macintosh in this test. For comparison’s sake, the quad-core 3GHz Mac Pro was 26 percent faster than the 2.5GHz quad-core G5.
[While this piece was being edited, Apple released both the OS X 10.4.10 upgrade and an update to its pro applications, including Final Cut Pro. A quick retest of the eight-core Mac Pro with 4GB of RAM found no difference in test scores.]
Speedmark 4.5 scores are relative to those of a 1.25GHz Mac mini, which is assigned a score of 100. Adobe Photoshop, Cinema 4D XL, iMovie, iTunes, and Finder scores are in minutes:seconds. All systems were running Mac OS X 10.4.9 with 1GB of RAM, with processor performance set to Highest in the Energy Saver preference pane when applicable. The Photoshop Suite test is a set of 14 scripted tasks using a 50MB file. Photoshop’s memory was set to 70 percent and History was set to Minimum. We recorded how long it took to render a scene in Cinema 4D XL. We used Compressor to encode a 6 minutes and 26 seconds long DV file using the DVD: Fastest Encode 120 minutes – 4:3 setting, using Apple’s Qmaster software to create a two-instance cluster to process the job. In iMovie, we applied the Aged video effect to a 1-minute movie. We converted 45 minutes of AAC audio files to MP3 using iTunes’ High Quality setting. We used Unreal Tournament 200′ Antalus Botmatch average-frames-per-second score; we tested at a resolution of 1,024 by 768 pixels at the Maximum setting with both audio and graphics enabled. We created a zip archive in the Finder from a 1GB folder. To compare Speedmark 4.5 scores for various Mac systems, visit our Apple Hardware Guide.
All tests are in minutes:seconds. All systems were running Mac OS X 10.4.9 with 4GB of RAM, with processor performance set to Highest in the Energy Saver preference pane when applicable. The Photoshop Suite test is a set of 14 scripted tasks using a 50MB file. Photoshop’s memory was set to 70 percent and History was set to Minimum. We recorded how long it took to render a scene in Cinema 4D XL. We used Compressor to encode a 6 minutes and 26 seconds long DV file using the DVD: Fastest Encode 120 minutes – 4:3 setting, using Apple’s Qmaster software to create a two-instance cluster to process the job. For the Pro App torture test, we recorded how long it took Photoshop to run our standard test suite while a longer Cinema 4D task and our Compressor encode test ran in the background.
All tests are in minutes:seconds. All systems were running Mac OS X 10.4.9 with 2GB of RAM, with processor performance set to Highest in the Energy Saver preference pane when applicable. The Photoshop Suite test is a set of 14 scripted tasks using a 50MB file. Photoshop’s memory was set to 70 percent and History was set to Minimum. We recorded how long it took to render a scene in Cinema 4D XL. We used Compressor to encode a 6 minutes and 26 seconds long DV file using the DVD: Fastest Encode 120 minutes – 4:3 setting, using Apple’s Qmaster software to create a two-instance cluster to process the job. For the Pro App torture test, we recorded how long it took Photoshop to run our standard test suite while a longer Cinema 4D task and our Compressor encode test ran in the background.
Buying advice
If time is money, upgrading to a Mac Pro from the last generation of PowerPC-based Mac towers probably makes good financial sense. But the eight-core 3GHz Mac Pro showed only very modest improvements in the majority of our testing over a standard configuration Mac Pro. Unless you’re an application developer looking to optimize your programs for multicore processing, or a user of the small handful of applications that are already multiprocessor aware, a quad-core Mac will probably give you more bang for your buck.
Comment: [email protected]
