Review home page | General Overview | Photoshop Elements | Games | Rob's Benchmarks | Other Benchmarks | Conclusion

Third Party Benchmark Tests

small logo

Strength in Numbers

In addition to my own homemade benchmarks, I ran a few more official benchmarks using freely available programs, including Cinebench 2003, XBench, SpeedRun, and a little app called Let1KWindowsBloom that does exactly that -- it opens 1,000 windows and reports how long it took to do so.

Cinebench 2003

From Maxon's site, "CINEBENCH 2003 is the free benchmarking tool for Windows and Mac OS based on the powerful 3D software CINEMA 4D R8. The tool is set to deliver accurate benchmarks by testing not only a computer's raw processing speed but also all other areas that affect system performance such as OpenGL, multithreading, multiprocessors and Intel's new HT Technology."

This seemed like a good way to get a feel for the difference between the G4 and the G5, and single vs. dual CPUs (the test runs both single and dual CPU tests). In the following table, bigger numbers mean better performance, except for the two test results shown in seconds -- in those two rows, fewer seconds indicates faster performance.

When I first ran the benchmark, I had the CPU set to "automatic" mode, so I ran the tests again with the setting on "highest," just to see what the difference might be -- both results are presented in the following table, next to the G4/733's figures.

 

As the table plainly shows, there's just no comparison between the two processors. Even on the single CPU tests, the speed increases are astonishing. But the real performance is demonstrated in the dual CPU tests -- what was a nearly seven-minute render takes just over a minute when done using both G5 CPUs -- and well under a minute with the G5 optimized version of the code!

XBench

XBench was one of the first benchmark programs for OS X, and it's relatively widely used. Once again, I forgot to set the CPU to highest before running the benchmark, so this table has the same three columns as the Cinebench table.

It's starting to sound repetitive, but the G5 easily trounces my G4 in every test across the board, and the ATI 9800 video card makes quick work of the OpenGL and Quartz rendering tests. My G4 was no slouch in the video department with its GeForce3 card, but the ATI simply leaves it in the dust.

SpeedRun

The SpeedRun benchmarks are relatively simple and run quickly. I have no idea as to their validity, but repeat runs generated results without too much variation, so there is some level of predictability. One cool aspect of the program is that it downloads a database of other machines' results for use in comparing results. This can help if you think your machine isn't performing up to its capabilities -- are your SpeedRun results similar to those of others with the same machine?

The G5 breezed through these tests without blinking an eye, and once again made the G4 feel much older than its two years.

Let1KWindowsBloom

As explained above, this app simply opens 1,000 windows and times how long it takes to do that task. If you visit the author's website, you'll see a fair bit of commentary as to whether this is a fair benchmark or not. Personally, I'm not sure I care about the validity of the results, only that they're repeatable from one machine to the next. If that's true, then you can still use the test to get a sense of the relative speed of one machine against another at that one task.

On my G4, it took 58 seconds to open the 1,000 windows. On the G5, it took only 18 seconds. Quite a difference!

Conclusion

I am not a professional benchmarker. I have no degree in statistics. I'm sure my control of the testing conditions is far from perfect. I'm not sure I'd want someone making a purchase decision solely on my test results. And yet, with all that out of the way, there's no denying that the Dual G5 turns in some simply astounding results on these benchmark tests.

But who really cares about benchmarks? At the end of the day, it's about how the machine actually works and feels, right? So read on to the Conclusion to get my final thoughts on that matter.

Review home page | General Overview | Photoshop Elements
Games | Rob's Benchmarks | Other Benchmarks | Conclusion

[Feedback is welcome -- please use this thread on our forum site to discuss this article.]

Contact Us | ©2003 macosxhints.com ... reproduction allowed with full attribution (including hyperlink) to source.