New quad-core and eight-core Mac Pro benchmarks

James Galbraith
30 August, 2010
View more articles fromthe author
AAA
Reviews

The new Mac Pros announced in July have finally arrived at at our testing lab. And while we ran into some hiccups during the tests, we’re able to share our first results.

The new Mac Pros use the identical case design as the systems they replace, with the biggest differences being in the available CPU and GPU options. Apple offers two standard configurations, a $3499 system with a single quad-core 2.8GHz Intel Xeon Nehalem processor, 3GB of RAM, 1TB hard drive, and ATI Radeon HD 5770 graphics with 1GB GDDR5 video RAM; and a $4899 system with identical graphics and storage specs, but with 6GB of RAM and two quad-core 2.4GHz Intel Xeon Westmere processors (eight cores total). A third system, a $6999 12-core model with two 2.66GHz six-core Xeon Westmere processors and the same storage, RAM, and graphics specifications as the eight-core model is considered a pre-configured build-to-order model, meaning you won’t find it on the shelves of the brick and mortar Apple Stores.

Compared to the only standard configuration offered previously – a system with a single quad-core 2.66GHz Intel Xeon Nehalem processor, 3GB of RAM, a 640GB hard drive and nVidia GeForce GT 120 graphics with 512 MB of GDDR3 memory – the new 2.8GHz system was 13 percent faster overall. Most tests were just a few seconds faster on the new Mac Pro, though there was a big gain in graphics performance, with the ATI Radion HD 5770 able to display 87.7 frames per second in our Call of Duty test, while the nVidia GT120 managed to display just 49.3 frames per second. iPhoto importing was 18 percent faster and iMovie exporting was nearly 30 percent faster.

Comparing the new eight-core 2.4GHz system to the new quad-core 2.8GHz system is a bit tougher because we can’t use Speedmark to compare overall performance. We ran into a problem running Parallels on the Westmere-based Mac Pros, so we could not complete the Speedmark 6 suite of tests. When booted into 64-bit mode (as the new Mac Pros do by default) the systems would crash with Parallels installed. Booting into 32-bit mode would allow the systems to start up, but we still couldn’t use Parallels. We eventually tried a complete wipe and reinstall from the Mac Pro’s systems disc and downloaded the very latest version of Parallels to eliminate any other potential problems, and had the same issue. Apple and Parallels are aware of the problem and assure me that Parallels is working diligently on a fix. The application worked just fine on the new quad-core Mac Pro with the Nehalem processor.

Looking at individual test results, we see that the eight-core system is much faster at the those higher-end applications that take full advantage of multiple cores. MathematicaMark, for example, scored nearly 44 percent higher on the eight-core system than the new four-core system. Cinebench was 28 percent faster on the eight-core system.

In many applications, however, having fewer, but faster processors was preferable; the quad-core 2.8GHz Mac Pro an edge over the eight-core 2.4GHz Mac Pro in tests like Aperture, iTunes, and Photoshop. Even our Compressor test showed the quad-core model to be 4 percent faster.

Comparing the individual scores of the new eight-core system to a build-to-order eight-core system from last generation that had two quad-core 2.26GHz Intel Xeon Nehalem processors, we see faster performance across the board, again with the biggest gain coming in our graphics tests.

Comparing the new quad-core 2.8GHz Mac Pro to the 27in 2.8GHz Core i5 iMac, we see the Mac Pro scoring almost 6 percent higher overall, with wins in Photoshop (10 percent), Cinebench and MathematicaMark (17 percent for both). Both iMovie tests and the Finder Unzip tests were also faster on the Mac Pro. The iMac finished the compressor test a little less than 3 percent faster than the quad-core Mac Pro and was 11 percent faster at the iTunes encoding tests.

Quad-core and 8-core Mac Pro (Mid 2010) benchmarks

Speedmark
6
Adobe
Photoshop
CS4
Cinebench
R10
Mathemetica-
Mark 7
Compressor
3.0.4
Aperture
2.1.4
iMovie
’09
iMovie
’09
iTunes
9
Call of
Duty 4
Finder Parallels
WorldBench 6
Handbrake
0.9.3
iPhoto
’09
OVERALL
SCORE
SUITE RENDER SCORE MPEG
ENCODE
IMPORT IMPORT
ARCHIVE
EXPORT MP3
ENCODE
FRAME
RATE
UNZIP
ARCHIVE
MULTI-TASK
TEST
RIP
DVD
CHAPTER
IMPORT
Mac Pro 2.8GHz quad-core
Intel Xeon (Nehalem)
230 0:36 0:57 11.8 5:47 1:34 1:03 0:55 1:05 87.7 0:38 4:13 1:14 0:22
Mac Pro 2.4GHz eight-core
Intel Xeon (Westmere)
NA 0:42 0:41 20.9 6:02 1:37 0:50 0:57 1:09 87.1 0:37 NA 1:10 0:22
Mac Pro 2.66GHz quad-core
Intel Xeon (Nehalem)
203 0:38 1:00 11.1 6:17 1:36 1:09 1:18 1:10 49.3 0:44 4:27 1:15 0:27
Mac Pro 2.26GHz eight-core
Intel Xeon (Nehalem)
206 0:42 0:45 19.5 6:33 1:45 0:57 1:24 1:15 50.4 0:41 5:00 1:10 0:28
27in iMac 2.8GHz
Core i5 quad core
217 0:40 1:09 10.1 5:36 1:47 1:17 0:57 0:58 88.1 0:44 4:03 1:01 0:22

Best results in bold. Reference systems in italics.

How we tested. Speedmark 6 scores are relative to those of a 2.13GHz Core 2 Duo MacBook, which is assigned a score of 100 (higher scores are better). Call of Duty score is in frames per second (higher is better). MathematicaMark is a performance score (higher is better). All others are in minutes:seconds (lower is better). New Mac Pros and the new iMac were tested with OS X 10.6.4. The iMac had 4GB of RAM, the quad-core Mac Pros had 3GB of RAM, the 8-core Mac Pros have 6GB of RAM. The older-generation Mac Pros were tested with OS X 10.6.1. We duplicated a 1GB file, created a Zip archive in the Finder from the two 1GB files and then unzipped it. We converted 90 minutes of AAC audio files to MP3 using iTunes’ High Quality setting. In iMovie ’09, we imported a camera archive and exported it to iTunes using the Mobile Devices setting. We ran a Timedemo at 1024-by-768 with 4X anti-aliasing on in Call of Duty 4. We imported 150 JPEGs into iPhoto ’09. 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 used Compressor to encode a .mov file to the application’s H.264 for video podcast setting. We ripped a DVD chapter to the hard drive. We recorded how long it took to render a scene with multiprocessors in Cinebench. We ran the Evaluate Notebook test in MathematicaMark 7. We ran the WorldBench 6 multitasking test on a Parallels 5 VM running Windows 7 Professional. We timed the import and thumbnail/preview creation time for 150 photos in Aperture.–Macworld Lab testing by James Galbraith and McKinley Noble

Check back soon for our full, mouse-rated review of the new Mac Pros. We’re also working on testing the new 12-core Mac Pro. I’ll also keep you updated on what I find out about Parallels compatibility with the eight- and 12-core Mac Pros.

Leave a Comment

Please keep your comments friendly on the topic.

Contact us