Anthony Frausto-Robledo, Editor ([email protected])

13 Apr 2001

Mac3D: G4 vs PIII shootout yields poor Mac G4 performance and controversy

I had learned from reader David Yuguchi that the March 2001 issue of Computer Graphics World magazine contains a 3D shootout between a Power Mac G4 and a Pentium III PC box. The Pentium III box reportedly toast the G4 in just about all test by 40 - 250 percent faster.

However, the test machines were comparable on cost (each at approx. $3000) not technical features. Apple provided Computer Graphics World with a Power Mac G4 500 dual processor machine with (and here is one problem) an ATI Rage 128 graphics card with 16MB of video RAM.

The PC was a Dell Pentium III 933MHz with an Elsa Gloria II graphics card with 64 MB or video RAM. The article was written by Joe Greco who happens to love the Mac, according to first hand knowledge. But the testing and assumptions were flawed in a number of areas. (see below)

Architosh Contacts Joe Greco on Report

I immediately contacted Joe Greco by email after I read the report. I had felt the report was unfair and I did not dispute the results (more on that below) but had a number of issues with how the test was performed. Joe then called me to discuss the matter further and explained his reasoning. Joe's report was done back in the Fall, which explains a lot of things and his intentions were not to disparage the Mac. Joe's a good guy who happens to still hold a special place in his heart for the Mac.

Some Major Misgivings

However, I suggested a different testing process not predicated on comparable cost. Here are the issues:

Cost Problems

Joe's test Mac had two processors in it but the test applications according to him did not take advantage of the second processor in the Mac, therefore rendering the value of that processor useless in this comparison. By paying for the extra but useless processor basing the whole article on 'performance-value' disadvantaged the Mac from the get go. It would be different if it was a Photoshop shootout.

Remedy: Don't base comparisons on cost to begin with. PC's just aren't equivalent to Macs in terms of features. Macs come with built-in FireWire ports, AirPort networking readiness, onboard Gigabit Ethernet, best-of-breed apps like iMovie and special Mac OS features like iTools, iDisk and HomePage which are all a part of the overall cost-value of the Mac. PC's don't ship with equivalent things, hence the cost way is the wrong way. And for professional 3D users, $500, even $1000, difference either way doesn't make a huge difference if the more expensive machine renders 20% faster -- it will pay for itself in time savings (billable hours) anyway.

Graphics Card Problems

One of the reasons why the Pentium III box beat the G4 machine was because of the graphics cards used. The ATI Rage 128 Pro is not a CAD/CAM class card at all. It's a multipurpose/gaming card. On top of that the nVidia-based Elsa card had 64 MB of RAM Vs the 16 MB on the G4's card.

Remedy: Use cards like the widely available ATI Radeon AGP card or nVidia GeForce 2 or 3 cards in both machines and match them up. This will also allow users to see the speed difference between the systems sans graphics card contributions. And this will also highlight the MHz myth as well.

Hard Drive Issues

Again, because the whole comparison was predicated on matching dollar values, the Dell Precision box came with a faster SCSI-based hard drive spinning at 10,000 rpm. This definitely offset the disk intensive tests in noticeable favor of the PC.

Remedy: Don't base the whole comparison on cost. Match up equipment to see how a Mac really stacks up against a comparable (not dollar equivalent) PC. Again, helping to keep certain equipment the same helps to isolate and compare the things that can't be the same (eg, processors, bus speeds, caches, OS, etc. -- things that make up the real differences anyway between Mac Vs PCs, between Apple Vs Intel/MS)

Conclusions

This could have been a really informative article. The only thing we learn in it is that for $3000 we can buy a Mac with a second processor that just sits there (but we pay Apple for it anyway) and run three Mac applications which don't utilize it and happen to respond much better to a faster hard drive, much more graphics RAM and more powerful graphics chip.

And to make matters worst, the average PC user will likely conclude that the G4 running at 500Mhz is the main culprit for the poor performance. If it had been running at 933MHz like the Pentium III maybe things would have been different. This just highlights the bogus MHz Myth and makes matters worse for Apple.

The good side of this story is that we understand now that cost-based comparisons are not very valuable, especially to professionals who charge high hourly rates for their time. For them they will make up the differences if a given machine works just 20% faster.

To read the Computer Graphics World article click here.