Overclocking

Overclocking was a mixed bag for us. Previously on the Abit and Asus boards, FSB speeds of 150+ were not uncommon. We worked in increments of 10, and here are our results...

12.5 x 140 = Pass
12.5 x 150 = Fail

Ok, so it looks like the multiplier may be adding too much strain to the system. Since we unlocked our XP 2000+ a while ago, we dropped the multiplier back to 10. We would have left it higher, but I wanted to just get a post...

10 x 140 = Pass
10 x 150 = Fail

This was getting interesting. I was fairly certain that 150FSB would be attainable, so we dropped down the multiplier to 9, which was the absolute lowest I would go without comprimising the performance too much...

9 x 140 = Pass
9 x 150 = Fail

At this point, I concluded that 150FSB, or higher was not going to happen. We set the multiplier back to it's default, and started playing with 1MHz increments. Just to keep you posted, we tried a whole bunch of settings, with limited success. Here are the results of the settings that worked...

12.5 x 143 = Pass
12 x 143 = Pass
11.5 x 143 = Pass
11 x 143 = Pass
10.5 x 145 = Pass (Windows lockup)
10.5 x 144 = Pass (3D Mark lockup)
10.5 x 143 = Pass
10 x 144 = Pass (Unstable)
10 x 143 = Pass

For the record, our ram was kept asyncronous with the FSB for testing. No matter what multiplier, at 146FSB, the system wouldn't even post. It powered on alright, but we were forced to reset the CMOS. At 144 and 145 FSB, we were able to post, sometimes run a benchmark or two, but the system was flakey. At 143FSB, the system was as stable as a rock. We pushed a lot of system stress tools, and it never wavered. There is a possibly of a BIOS setting we missed, but no matter how hard I tried, it just wasn't happening at 144+.

I've been speaking with a few other webmasters about that issue, and I've been told by a few that the KT333 revision (we had the "CD" one) may be the culprit. The "CE" revision is reported to overclock much higher.

Test Setup

AMD Athlon XP 2000+
MSI KT3 Ultra-ARU
512MB PC2700 Crucial DDR
2 x 60GB 7200rpm Maxtor Harddrives, Promise RAID
Visiontek Xtasy GeForce 4 Ti4600

AMD Athlon XP 2000+
Asus A7V266-E
512MB PC2100 Kingston DDR
2 x 60GB 7200rpm Maxtor Harddrives, Promise RAID
Visiontek Xtasy GeForce 4 Ti4600

Windows XP Professional
Via 4-in-1 v4.38
nVidia Detonator 28.32

SiSoft Sandra
PC Mark 2002
Quake 3: Arena
3D Mark 2001SE

Rather than bogging you down with a dozen benchmarks at different speeds and resolutions, we're going to keep things simple. When applicable, only 640x480 scores will be displayed, as that resolution will eliminate the video card as the bottleneck.

Since this is the first KT333 board we've tested, we're going to compare it against the fastest KT266A we've tested so far, the Asus A7V266E. Because of the limited upgrade from the KT266A to the KT333, we feel the comparison will be fair, and reflect whatever improvements VIA may have done. For our memory benchmarks, all setting were configured for maximum performance. We did have to back down to CAS2.5 for 166FSB+ for our tests to complete successfully.

Previous Page - The BIOS

Next Page - Benchmarks