Tulatin · 08-26-2005 · Category:
So you've read through the procedures, the clocks and the hurdles - we bet you want to see how the new, higher clocks of the processor will improve over the old ones. We've decided to do a full gamut of benchmarks here, to show just how large of improvements can be made across the board.
We've decided to start things off simply, utilizing Si-Soft Sandra to show theoretical improvements of CPU Power and Memory Bandwidth. Next, we'll utilize Super-Pi at levels of 1, 4 and 8 Million places to show the real world effects of the boosts which Si-Soft Sandra told us about. Along with this, we've decided to encode a full length film (in this case American Pie) to Divx with AutoGK.
With the Trio of CPU and Memory tests aside, we've decided to see how much of an effect these boosts exhibits in gaming scenarios - first theoretically, then actually. We begin this little experiment with 3DMark 2001 and '05, with the former being heavily influenced by CPU power and the latter requiring considerable amounts of physics computation.
In the end, I definitely like the results I'm seeing here. By providing a quaint 25% increase in processor and FSB frequencies, we've achieved an average of 13% increased processing power and bandwidth, along with a 15% average increase in 3D Mark Scores (albeit, with little of this average being shown by 3D Mark 2005). While the increases in 3D Mark scores are nothing earth shattering, the improvements in processor performance are - after all, taking 12 minutes off a DivX encode, and 10 seconds off of a Super-Pi run are definitely significant, and are likely to increase as the processor's speed scales up and the FSB opens wide.
- The Hardware