...I have to admit I was surprised. In apps that are memory bound, the 655TX appears to be faster than Canterwood. The Springdale-PE on the DFI, backed up by a capable PAT BIOS means that we were emulating Canterwood. It has lower memory access latency, measured by Sciencemark, and it appears to have a touch more available bandwidth too. It's not much, but it's enough. With Canterwood the performance champion for P4, it appears we have a new champion when your app is memory bound in any way. Raw CPU performance is identical; 655TX ran the 3.2GHz Northwood without issue.