First let's look at our test rig used for all benchmarks:

Test configuration
Motherboard	MSI K9N SLI (nForce 570 SLI chipset)
Processor	AMD Athlon 64 X2 6000+ (2x 3.0GHz, 2x 1MB cache)
Memory	4GB: 4x 1GB DDR2/666 4-5-4-15
Storage controller	onboard nVidia nForce 570 Serial ATA controller
Video card	GeForce 8600GT 256MB DDR3
System disk	Hitachi 160GB SATA (HDT722516DLA380)
Operating System	Windows XP Professional SP2 & FreeSBIE 1.1
Drivers	Forceware v162.18 WHQL, AMD CPU v1.32 WHQL, nVidia ATA v5.10.2600.0666 WHQL

Careful benchmarking
Due to the nature of the issues I know I had to perform very thorough and strict benchmarks. I made sure the scores were consistent and ran each benchmark at least 5 times, where the average was taken as final score. If a benchmark score clearly deviated from earlier scores, another benchmark run was done to replace its score. I had to do this twice with the WinRAR tests.

In order to restrict the influence of the significant amount of RAM, I used RAMdisks which consumed close to 3GB of memory. The benchmarks were run in a pristine state, as far as possible on Windows, meaning:
All non-relevant applications killed off using task manager
Stopping system services like: automatic updates, indexing service and security center
Disconnecting the computer from the network

I even went as far as to kill off explorer - I really wanted to make sure nothing on the system could influence my scores. I wanted the disks to sweat and know it's crunch time.