About The Benchmarks
The benchmarks proved to be very interesting. The unlocked AMD Athlon Thunderbird 1.33 GHz CPU was run at its default speed (133 X 10) as well as 1411 MHz (166 X 8.5) and 1485 MHz (166 X 9). The RAM was a pair of Mushkin PC2700 256MB, for a total of 512MB. The video card was an Asus GeForce 2 GTS 32MB DDR. The hard drive was a Maxtor ATA133 drive running Windows XP.
3DMark2001 would not run at 1485. This was certainly due to the CPU. Even with the voltage increased by 10% and a Coolermaster Heat Pipe, 3DMark2001 would not run at this speed or beyond.
Evolva saw an increase of minimum FPS at 1411 and an increase of maximum FPS at 1485. The bumpmapped benchmark, however, showed a slower average and maximum benchmark with the 166 MHz FSB. We thought that there was actually a very good reason for this. We had thought that at 166 MHz FSB, the AGP was only running at 55.3 MHz as opposed to 66 MHz due to a 1/3 divider instead of a 1/2. Iwill assures me that the AGP divider is 2/5 at 166 MHz FSB.
Unfortunately, I could not prove or disprove this. Ordinarily, I would go into SiSoft Sandra and click on "Mainboard Information" and scroll down to "AGP Bus". For some reason, when I tried this on the Iwill XP333R, there was no AGP bus information!
I'll just have to take Iwill's word for it about the AGP divider and figure out why the slower benchmarks at 166 MHz at another time.
We had thought that perhaps Evolva was just being a crack-head, so we tried Quake 3 Arena.
We all had a good laugh over the Q3A benchmarks. After establishing a base with the CPU running at 1333 MHz with it's default 133 MHz FSB, we stepped up to 1411 and we obtained nearly identical scores!
Certainly, the GeForce 2 being used is a year shy of being state of the art, so it was speculated that the reason for the Quake results was a bottleneck at the video card. We then tried 1485 MHz and Q3A locked up like 3DMark2001 did. Can I get a collective Homer Simpson style "DOH!"?
The Sandra scores are all self-explanatory. The faster the CPU, the better the CPU benchmarks. And of course, the memory benchmarks went up as the memory's front side bus went up. The only time the memory benchmarks did not go up when the memory's bus went up was when we tried to run the RAM's speed asynchronously with the CPU's FSB. But as stated earlier, this is something Iwill is working on and is currently just a flaw in the current BIOS. For this reason, I did not even post the memory benchmarks in this review, as it would be negative press that simply has no bearing on the actual performance of the board. What these benchmarks will be AFTER Iwill fixes the BIOS is something that's worth reporting, and if I remember to do so, I will post an update with the asynchronous scores at that time.