Installed the FPR 3 and so far, an entire day in use, it remains much more responsive that FPR 4.
All iterations of TFF slow down eventually and need quitting/restart (I tend to keep open several windows at the same time and probably this is the culprit), but FPR 4 was unusually quick - just a few hours in use - in becoming unresponsive.
The slowest loading site I ever came across is Google Maps: sometimes, when TFF gets bogged down and needs quitting/restart, Google Maps page download only partially and then freeze; or download it its entirety, but doesn't respond anymore.
Yes, TenFourFox certainly does accumulate cruft after awhile (this is not a TenFourFox-specific problem; regular Firefox also does this). Multiple open windows and tabs can make it worse, though you'd probably need open windows in the double digit range to be the cause. Some of the local changes I've made were to make garbage collection more deterministic but there were no such changes between FPR3 and FPR4. For the record, on the G5 I'm typing this on, I have about 7 tabs active and 43 tabs idle and it has been running since November 9 (i.e., 17 days of the browser running continuously without sleeping or restarting). I don't have any beachballs or slowdowns currently as I type this.
I did try both sites Ron suggested, but I'm unable to find a quantifiable difference between FPR3 and FPR4 on the test systems. It's possible your systems are configured differently to mine, of course. I didn't receive any reports like this during the testing phase of FPR4, and at least some feedback that the browser was noticeably faster, so I can't clearly determine exactly what's going on.
That said, I'll leave this ticket open to see if there are other reports or if you discover something additional. If there is a site that is consistently problematic and is clearly better in FPR4 than FPR3, though, that could be helpful to know (but again, I'd need to reproduce it in a controlled environment to be able to do anything about it).