Once again I doubt many around here have experience with serious video machines (I doubt I can really count myself in this bracket either). Can I assume this is both producing and encoding/authoring?
First: I encode/remaster/recut and occasionally produce a hell of a lot of video ranging from mpeg1 through DVD through MPEG4-ASP to MPEG4-AVC (any and all containers really). I rarely, if ever, do WMV or other easily streamable stuff (not that it really makes much difference here). I do not however use the apps you do (the GIMP and avisynth are my comparative choice for those).
It also must be said the CPU is 99% of the time the limiting factor (capturing is hard drive limited normally and once in what was almost an experiment (simultaneous encoding of 2 heavily filtered (mvtools+spatio temporal stuff) high res videos with a fairly early version of X264 I ran out of ram)). You mentioned once something to do with an encoder card, if this is the case I may have to revise my advice although the fact it is not here probably says something. Same goes for graphics cards enhanced encoding (although normally that means a really nasty codec is made or the ability is hidden behind an obtuse/rarely used (outside of tech demos) API).
I know you said no overclocking but it should be said with a simple FSB overclock (which these core whatever chips seem more than capable of: I have overclocked about 6 core2 and one quad through the roof on standard cooling and I am nowhere near the levels of some of the overclockers) you will probably get nearly the same stable clock speeds out of them, having participated in several of these threads over the last few years now I doubt you will have any trouble with "reduced life".
As for quad not being of any use: for games and general UI I would tend to agree however in the simplest form multicore encoding for me means splitting the video in half and encoding, for which there are now several applications (although normally using avisynth as a frontend, occasionally they just join though), you may get a 200 KB or so increase in size for a DVD but that is not really here or there.
If manually authoring a DVD I tend to encode each tile individually but they are normally the same length and level of complexity or occasionally I might keep the another core for audio.
I realise this probably has not said much so in my opinion it really boils down to this:
does you app support multicore (even with a kludge like encoding to lossless (can I suggest you try h264-lossless for this if you have not already) and then transcoding or 30 minutes on a reusable batch file)? If yes quad, if no (or you really consider it unreasonable) clockspeed is the limiting factor in encoding most of the time (a small cache which is not an issue here can also mean a noticeable hit) the the higher clocked may be a better option. My personal preference is towards the quad though and it really is not difficult to set up multi core stuff.
Although I appreciate this is a here and now thing SSE5 (which could herald some nice improvements in this arena although it will take a while for support in encoders) as well as AMD offerings:
http://www.guru3d.com/newsitem.php?id=5818
