The Computer Blog

Friday, May 23, 2008

Barefeats Was Right!

It was Rob-art’s testing at his Mac performance website that alerted me to the problems with the NVIDIA 8800GS GPU chipset being used in Apple’s iMac. I keep a Mac Pro for really heavy graphics and video work and gaming, so my primary use for an iMac is to work with “everyday applications” and some light photography and video work. When I was ordering it some ten days or so ago, I decided to order a 3.06 GHz iMac with the Radeon 2600 XT GPU rather than the NVIDIA 8800GS GPU, not only to save me $150 but also to optimize the machine for the things I’d be using it for. Of course, I knew there was some possibility that updated drivers might improve the NVIDIA’s chip’s performance, but something in my gut told me that Rob is probably right about there being a fundamental flaw in the NVIDIA GPU when it comes to handling Core Image functions. I have to admit, too, I have a personal bias toward ATI chips, a holdover from my PC building days. I really love them, and I try to buy systems with them whenever I can.

Last night, I finally got the time to run a couple of the tests Rob-art used in his evaluation of the Radeon 2600 XT versus the NVIDIA 8800GS. When I ran the iMovie 7 Import test, the import time came out to be exactly the same as his (running the last generation’s 2.8 GHz iMac with the Radeon 2600), i.e., 40 seconds. While that was a bit of a personal disappointment, it showed that the GPU had more influence over the iMovie 7 import task than the CPU and that the Radeon is definitely the better GPU for the job. The extra speed of the 3.06GHz CPU came more into play when I performed the iDVD Encoding task. There, the task completed in 135 seconds versus Rob’s 147 seconds using the 2.8 GHz iMac running the same GPU. Both were significantly faster than the 3.06 GHz iMac running the NVIDIA 8800, as you can see here.

By the end of the holiday weekend, I hope to have run benchmarks using Cinebench 10, the iMovie 7 Import test, the iDVD Encode test, and perhaps one or two more test using my 3.06 GHz iMac, my 2.66 Mac Pro with its stock NVIDIA 7300 GT video card, and my 2.33 GHz MacBook Pro running its Radeon X1600 with 256MB, as well as my wife’s 2.0GHz Core 2 Duo AL iMac and her 2.0 GHz Core Duo MacBook. (I’d like to run Photoshop CS3 benchmarking tests, but the recently discovered 20 authorization/deauthorization limit for CS3 products make me hesitant to move a copy of CS3 around.) I will discuss the results here and probably also post them separately on my website to make them readily available without having to search through the blog. I am looking to buy the Radeon 3870 HD and run it in my Mac Pro when it’s released and will add its benchmarks to the mix as soon as I can. (Rob-art will have a more complete analysis on the 3870 and probably one up sooner than I will, so check there first if you’re interested in that. ) I’d especially love to tie in some Cinebench testing with the G5, G4, and AMD shootouts still posted on my site, but I’m not sure I’ve got a common version of Cinebench to run. That said, some of the machines benchmarked in those old articles are still in my family and most can run Cinebench 10 if I can get the owners to run it for me and send me the data.

It’ll be interesting to see how it all turns out and what I can put together, so I hope you’ll drop by every now and then to see what’s going on.


Post a Comment

<< Home