The Computer Blog

Monday, June 09, 2008

Benchmarking 4 Macs – iLife and Imaginator Benchmarks

If you missed out on the first blog about these benchmarks, then I’ll share again that I’m going this primarily because I bought a 3.06 GHz iMac with the Radeon 2600 HD GPU. I wanted to be able to contrast its performance with “standard” 3 GHz iMacs running the Nvidia 8800 GS and also against the other Macs in my personal stable, none of which are more than a couple of years old. Since I started that task, the 10.5.3 update has arrived; and that, in and of itself, is producing some interesting results.
For the first set on benchmarks, I used Cinebench 10 as my major benchmarking tool. For this set, I used two iLife applications and Imaginator, a third party application built to pull specifically on OS X’s Core Image graphics engine. (Thanks to Rob-art at Barefeats.com for coming up with these tests.) Let’s take a look at the iLife tests first.

The first test utilizes iMovie 7 to import a movie. The movie I used was named “Wildlife” and is a high-definition demo movie that can be found at Apple’s Quicktime website. As iMovie import the other movie, it renders thumbnails of its scenes to use in the editing process. The chart below shows the results of the test: (Less is Better.)

The most surprising result of this test was that both iMacs beat my 2.66 GHz Mac Pro. Considering the differences in CPU clock speed, this suggests that the iMovie Import process is strongly affected by GPU performance. The GPU’s in the iMacs are ATI Radeon 2600 “Pro” in the 3.06 GHz machine and the Radeon 2400 in the 2 GHz iMac. The MacBook Pro uses an ATI 1600 XT Radeon. The Mac Pro is the only machine using a Nvidia GPU and not a particularly strong one at that, i.e., GeForce 7300 GT. It appears that the recent ATI Radeon GPU’s exhibit significant performance increases over ATI Radeon chips only a generation or so back.

The other interesting thing to note is that the 10.5.3 update seems to have had a slightly positive impact on this benchmark. While the times on the 2 GHz iMac and the MacBook Pro did not change, the Mac Pro and the 3 GHz iMac were slightly faster on 10.5.3. Barefeats.com established that there is a fairly significant speed-up for those users with the Nvidia 8800 GS GPU, though the Radeon 2600 GPU on the same machine still beats the Nvidia for this task. But what happens if we move on to encoding a DVD?

The iDVD Encoding Test uses the same “wildife.mov” file as the previous test to create a disc image of a DVD that uses the movie for both its opening movies and the movie file itself. The results I got are charted below:

Now, here’s where things get really funky. Under 10.5.2, the machines line up in the order you might expect based on CPU cores and speed. The 2.66 Mac Pro is the fastest, followed by the 3 GHz iMac, the 2.33 GHz MacBook Pro, and then the 2 GHz iMac. Something unexplainable happens under 10.5.3. The Mac Pro with its Nvidia GeForce 7300 GT GPU slows WAY down as does the ATI Radeon 1600XT on the MacBook Pro. Yet, the two iMacs with Radeon 2400 and 2600 GPU’s speed up! This suggests that the update is strongly tweaked for performance for the last two generations of machines. Older machines may actually slow down with this update. That’s not the way to go!
The test using Imaginator times how long it takes a Mac to morph a picture of a leopard into a rose. As I mentioned earlier, this application pulls strongly on OS X’s Core Image functionality. The results of that test are shown below:

What’s surprising about this test is the relative slowness of the 2.33 GHz MacBook Pro and the relative speed of the 2 GHz iMac. If I largely ignore CPU speed, then the test shows that the Radeon 1600 XT is inferior to the Nvidia GeForce 7300GT which are both inferior to the Radeon 2400 and Radeon 2600 GPU’s respectively. The only other reason I can think of that the MBP would be so much slower would be due to memory allocation and paging, but that seems really unlikely since the machine is running 3 GB of RAM and I observed nothing to indicate that was happening during the testing.
Clearly, all these tests indicate that the current batch of iMacs is fast enough for use in almost any environment, including those producing graphics and video. That’s not to say that they are the right choice for a heavy production environment. The Mac Pro’s additional speed (8 core) and its expandability still make it more suitable for that; but for a small business or prosumer, making the choice between those two machines are now much harder.

For me personally, the tests have pointed out the weakness of my Mac Pro’s stock Nvidia GeForce 7300 GT video card. I plan on replacing it with a Radeon 3870 HD card as soon as it is released, which will hopefully be soon. Once that’s in, I’ll rerun the tests to provide a comparison performance between “then” and “now”.

I’d like to close this out by switching gears only slightly and talking about the 2.8 GHz iMac sitting on the shelves now in a nearby Apple Store. As part of the examination of the Radeon vs Nvidia phenomena, I went to a nearby Apple Store and ran the iDVD Encode test using a 2.8 iMac. Much to my surprise, it ran the iDVD encoding test (under 10.3.5) in 215 seconds, about 13 seconds faster than my 3.06!

Unfortunately, I only had time to run one test; but just that one tells me you won’t lose much if anything by buying the 2.8 GHz machine. Secondly, testing at Barefeats is showing that the 10.5.3 update has narrowed the gap for Core Image functions if you’re running the Nvidia 8800 GS GPU. You can’t really lose by buying either of the speedy 24 inch iMacs available.

0 Comments:

Post a Comment

<< Home