Subjective Performance:

Subjectively, this board performs great. OS loads very fast, windows runs very smooth, and applications feel very snappy. While this is non-empirical evidence and just my personal opinion, I can say that I have used hundreds of mainboards including six or so Socket 939 boards and this is among the very best in how the "feeling of speed" is perceived by me. To qualify this statement, I have used boards that have benched very well in the past but felt sluggish when navigating windows and or running applications. This is just my opinion and a purchase decision should not be based off this alone. So let’s move onto the benchmarks.

Benchmarks

Sciencemark Beta 2.0

Aida32 Read/Write Tests

SuperPi 1mb test(lower is better)

SiSoft Sandra Unbuffered Memory Tests

3DMark2001SE

Doom3-run at 640x480(Low quality)

Benchmarks were run at default 200 Mhz HTT, 217HTT, 240HTT, and 265HTT to show performance levels when scaling HTT bus while at the same time keeping overall cpu speed at or near 2.4ghz.

The CPU multiplier was adjusted to accomplish this by setting multiplier to 12x for 200HTT(2400mhz), 11x for 217htt(2387mhz), 10x for 240HTT(2400mhz), and 9x for 265HTT(2385mhz).

The beauty of AMD Athlon64 chips is their downward unlocked multiplier on standard CPUs, and fully unlocked multipliers on FX series chips. Therefore, any potential benefit seen in higher HTT bus can be entirely attributed to the raise in HTT and consequently ram speed versus an increase in CPU MHz as would be the case with an Intel setup. Doom3 is run at low resolution and detail to more carefully show how higher HTT bus speeds effect performance as at higher resolutions the graphics card becomes the bottleneck more so than the mainboard / memory subsystem.

Results:

Sciencemark 2.0 Beta (mb/sec)
200HTT	217HTT	240HTT	265HTT
5573.91	5887.14	6444.29	6735.85

ScienceMark begins a trend I think that will be familiar throughout testing. We see that memory subsystem scores improve steadily as HTT speeds and in addition ram speeds are scaled. Quite impressive is the 265HTT score pulling 6735.85mb/sec, very high bandwidth in my opinion.

AIDA32 Read (mb/sec)
200HTT	217HTT	240HTT	265HTT
5666	5962	6543	6950

Aida32 Read tests are very similar to the ScienceMark runs. As ram and HTT speeds scale everything improves 22% at 265HTT compared to 200HTT.

AIDA32 Write (mb/sec)
200HTT	217HTT	240HTT	265HTT
2339	2435	2671	2950

Aida32 Write shows the potential of higher writes speeds as ram and HTT speeds scale. A 26% improvement in write speed is gained by going from 200 to 265HTT and Ram speeds.

SuperPi 1mb test (seconds - lower is better) (mb/sec)
200HTT	217HTT	240HTT	265HTT
36	35	35	34

SiSoft Sandra Memory Scores (mb/sec)
	200HTT	217HTT	240HTT	265HTT
Int Buff'd	6064	6416	6703	6946
Float Buff'd	6023	6345	6609	6864

No surprises here. Sandra continues the trend. Higher HTT and ram speeds shows marked improvement as buses increase.

3DMark2001SE Build 330
200HTT	217HTT	240HTT	265HTT
25081	25325	25598	26016

Here we see 3D advantages of higher Ram and HTT bus speeds. Steady improvements as we go higher. Improving almost 1000pts is nothing to sneeze at. Some people spend hundreds of dollars on a faster CPU speed grade, or better ram to allow gains like this. While not as dramatic as a next generation video card, a near 1000pt gain is definitely nothing to sneeze at. Impressive in my opinion!

Doom3 640x480 Low Detail (fps)
200HTT	217HTT	240HTT	265HTT
105.4	107.1	109.1	111.6

Doom3 was arguably one of the most anticipated titles of all time. Combining sheer terror with jaw dropping graphics this game oozes quality in graphics, sound and game play. A byproduct of such an extreme jump in game detail requires much higher system specifications than past games. When running this game at low detail and 640x480 resolution we get lower framerates than most newer systems with similar graphics cards get running an older title such as Quake3 at 1280x1024x32 and high details. Such is progress and the eye candy is certainly appreciated. That said, frame rates increase steadily as Ram and HTT speed increases. At higher resolutions the gains would be much smaller as the graphics card would be our limiting factor, but at these lower resolutions we can better illustrate the potential performance gains to be had by running higher Ram and HTT speeds.

Onto the conclusion ->