I'm not gonna get in an arguement over this, but I have a JVC D303 that's rated at 600 lines horizontal. I can definetly tell the difference between 640*480 and 800*600. The 800*600 is not as clear because technically, obviously, it's dropping pixels out of necessity. 1024 is even more painful on the eyes (especially if you need to be able to read it). If you have HDTV that's another story (But then you can get ATI's awesome compent dongle guaranteed). But the easiest way to see it is to open a web page and try reading at 640, 800, and 1024. The higher the res. (And the more data that's being dropped), the harder it is to read. 640 is the magic number because on a decent analog set, little to no pixels are being dropped.
The issue here, if I'm not mistaken, is that S-Video can only output "standard" TV resolution, so no matter what you crank it up to, it gets scaled back down to NTSC, or PAL, or whatever your local standard is. This is even if you are outputting to an HDTV! As long as you are using S-Video, you will always get scaled down to standard TV resolution.
In theory, a DVI connection will support higher resolutions if you are using an HDTV, but I haven't actually tried it. Component video also supports higher resolutions, but I've never seen a video card with component video on it (but I imagine they exist).
I don't believe this is a limitation of the cable, but rather just how the interface is implemented. On my HDTV, I have my DVD player plugged in via S-Video, and also using component. The component version is clearly taking advantage of 480p (an HDTV standard), while the S-Video connection is clearly using NTSC (which looks like crap on a big screen).
By the way, everything I've said here about S-Video also applies to standard composite video as well (but it looks even worse than S-Video).
-Jeff "etumor"