by Tim "Flyboy" Henderson

  I was doubtful about this estimate until Viking1 spoke to a UK sim developer, Andrew Walrond of Wayward Design. Apparently early code of B17 II simply flies along on the Athlon, much faster than on the Pentium III. The reason? Floating point performance. Andrew also mentioned that optimization of the code will eventually bring Intel performance more in line with the Athlon.

The Athlon in Benchmarks

Never one to rely on second hand information for more than a year or two, I was anxious to complete my own tests. Subjectively, firing up WIN98 second release on this system was much faster than my aging PII 450 (VBG). Opening and closing WORD and Outlook was also faster. Heck, even NHL 2000 played better! Oh.. you want to know about sims?

I'll get to that. I've also been playing with the Treemark test that Nvidia released as part of their GeForce 256 promotion. This test makes any current state-of-the-art system look like something from the ice age. (Get it here or from Nvidia.)

Now, I wasn't born yesterday. Obviously, the TreeMark test (simple version) is polygon intensive and designed for a dedicated T&L engine. But games are becoming more complex, and the trend is going to continue. So, it's likely a floating point biased, Arnold Schwarzenegger Conan type killer for any system around today. I thought it would be fun to run some comparisons. Heck.. it was! Here are the results on my system as well as Viking1's AMD K6-3.

AMD K6-3 450 TNT2 Ultra

• frames per second: 4.54814
• Screen Resolution 1024x768
• Frame count is 1000
• polygons per frame: 35820

PII 450 Millenium G400 MAX

• frames per second: 5.99125
• Screen Resolution 1024x768
• Frame count is 1000
• polygons per frame: 35820

AMD K7 600 Millenium G400 MAX

• frames per second: 9.00901
• Screen Resolution 1024x768
• Frame count is 1000
• polygons per frame: 35820

Click to continue

The TreeMark shows a large jump from the floating point challenged K6-3 to the Pentium II. But moving from the PII 450 to the Athlon 600 shows a 50% increase! It's a huge difference, and confirms the horsepower of the Athlon FPU.

Falcon 4 and WW2 Fighters

On to the real world applications, in this case the two simulations that can bring a PC to its knees and also include frame rate counters. Granted these are not scientific benchmarks, but at least you can know that the tests were carefully controlled and consistent across these three machines.

The TE tests were run in a TE campaign created by Viking1, flying an early HAVCAP mission with a single wingman. The environment was active with an ongoing ground war and many flights in the air. All tests were with full realism, clouds ON and texture smoothing ON. Both lift line and reflections were also ON, and BUBBLE set to 3. All other object and detail settings were untouched and identical for all tests on all boards.

F4 1024 2d cockpit.

Actual measurements in the TE mission for the G400 MAX resulted in 53.5 fps for the no cockpit view on the PII 500, and 58.8 fps on the Athlon 600. All settings were identical at 1024x768 resolution. (Thanks to the Ed for building the charts for me!)

Go to Part III: WW2 Fighters