The Many OS software tweaks of SuperPi Exposed

Over the years there has always been various tweaks touted and spread over overclocking forums on how to improve benchmark results. One area this has been quite prevalent in is SuperPi. We see one guy try something different, get a faster time and suddenly everyone's trying it for themselves. Too often we simply try the newest thing with no real understanding of what it is doing and how it is supposed to help our times. Then when we don't see any real gains we are left wondering what all the fuss is about or why it works for some people and not us.

What I propose here is a thread in which we can all input our findings to test these various tweaks. Each week we shall choose one tweak - specify a platform and run various SuperPi sizes (1M, 2M...) and hopefully through numbers we will establish what gains if any that tweak brings.

To begin with I ask for you to suggest any additional known software tweaks that are commonly used. I have made a list of the ones I know of so far but am welcome to suggestions as we go on.

Known tweaks

• XP or 2003 Server
• Copy Wazza
• LargeSystemCache
• /MAXMEM or /BURNMEMORY
• GUI Theme
• Processor Affinity and Priority
• Background Services
• RAMDISK Applications

Scope for Testing and Limitations

While testing these various tweaks I was confronted with a number of realisations.

• Different platforms and indeed processors seem to be affected by these tweaks in different ways
• Platform Chipset vs Tweak effect
• Physical RAM size vs Tweak effects
• Is there the possibility that the tweaks will interfere with one another - I tested them individually but what happens when they are combined?
• How many runs can be deemed as conclusive?