ZingDash

3DMark01

Looking at the figures shown one could ask, is all the tweaking worth it? Benchmarking performance with the various configurations of FSB/divider and memory timings is remarkably close. For gaming systems it is perfectly acceptable to use around 400-430 FSB on the 266 strap, as frame rates are not adversely affected by the FSB deficit.

Obviously those who enjoy benchmarking will prefer the potential of higher bandwidth, as sensitive benchmarks like SuperPi 32M and 3DMark 2001SE can take advantage of the extra tweaks the board offers to maintain low access latency along with high FSB speeds. Running high FSBs on last generation chipsets usually resulted in fairly apparent performance losses, which at times could even be overcome with lower CPU speeds while using a tighter chipset strap. DFI's implementation of the P35 chipset maintains bandwidth scaling if the relevant BIOS tweaks are applied.

It is clear that modern gaming engines are not as dependent on outright CPU/RAM performance tuning, certainly not to the level many of us tweakers would like to believe. The most important factor with Intel Core 2 processors still revolves around the 266/333 straps and actual CPU MHz speed for noticeable gains in overall gaming performance. Luckily enough, those using Q6600 CPUs have the option of the 9x multiplier, allowing us to overclock to the limits of current air/water-cooling capacity quite easily without the need to spend large amounts of time chasing higher FSBs of little visible gain.

Another positive here is that 4GB overclocking with the right modules is actually a lot easier than we are accustomed to on previous chipsets. Attaining higher RAM speeds with tighter sub-timings almost feels akin to overclocking 2GB of RAM rather than 4GB. Obviously XP 32-bit will not take full advantage of the additional 2GB of RAM. Vista 64-bit testing is underway to study the variance between how the board reacts to a true 4GB load.

ncG1vNJzZmivp6x7orrAp5utnZOde6S7zGiqoaenZH90gZNoaHE%3D