My efficiency sucks compared to what Thocathe and DopeLex show on the same CPU

I'm not sure if board (P5E64-WS Evo) or memory (Corsair GTX2) affects anything in wPrime, I simply kept the settings that I ran Pi and PiFast at. 738xx is perfectly reproducable with this exact combination of CPU/memory multipliers and strap/memory settings. I was previously around 74100-74200 for wprime 32M on Wolfdale 2M but I recently discovered that using (much) more wPrime threads than two (there is one exact value that worked best for all my 775 cpus) helps in cases where individual core completion times are far apart (by splitting the computation in smaller chunks thus leveling the load).