Please note that this might not be all TDE's fault. I have noticed that the X server (and possibly the kernel itself) tends to get slower and slower from release to release on old hardware. In general, locking myself to an old version of the kernel and Xorg on old hardware, then compiling new software on top of those old versions, seems to give halfway decent results.
I agree the problem is not TDE per se. I too seem to believe that Linux based systems get slower with each new release.
However, if you are noticing that TDE is running slower than KDE 3.5.10 on the same X/kernel versions, then we have a problem. ;-)
I haven't tried such a test on my older systems. On newer Linux systems, compiling 3.5.10 now is all but impossible with all of the various upstream software changes that require patching. I experienced that compiling R14 on older Linux systems is impossible without modifying or updating several distro packages. That limits any 3.5.10->R14 comparison to a specific period of Linux OSs. Fortunately, I have a candidate here (I can compile 3.5.10 and R14 on Slackware 13.1) and will give this a test a go.
That said what kinds of system or usability tests would be representative?
Darrell