sigh..
a Duron isn't really different, besides its cache, and the materials that it is made of.
really.
AMD produces Durons at it's Austin FAB at .18 micron without Copper (AFAIK), which limits it currently to about 1 ghz at best.
AMD makes their T-Birds (known as Athlons) in the Dresden fab, at .18 micron, and using copper interconnects, which allows it to (currently) hit a max of 1.4 or so ghz.
the K6-3+ cannot even outperform a Duron in GAMES, but probably can in Integer intensive apps, simply because it's got GOBS of cache behind it, and given that the other CPU's aren't clocked much higher then it. the K6-3 outperformed alot of the competition in that respect when it was in its day.
the K6-3 line is derived from the K6-2, execpt it has 256k ONBOARD L2 cache, whereas the K6-2 doesn't have any onboard L2 cache.
the K6-x series are based on the K6 CPUs, except have a new instruction set called 3DNow!, which in most cases doubles it's performance in FPU intensive applications, when implimented into software (ie, the software has to support 3DNow!, and use it PROPERLY). a good example of this was Quake 2, to which the boost was phenominal, becuase it was implimented very well. in todays games, the boost would be not quite as good, but still noticeable. too bad developers do not really impliment it, for us poor K6-x owners!
in fact, the K6-2 line very nearly equalled the Intel Celerons (In FPU intensive applications, which is what 3DNow! was made to do) it was competing with, however, then Intel came out with the smash Celeron A, which had onboard 128k cache, which hit AMD where it hurt, becuase their CPU didn't have any onboard L2 cache, thus giving the Celeron A, a new lease on life (in games compared to the K6-2).
AMD introduced the K6-3 somewhere after the CeleronA was released (I think). it had the 256k ondie L2 cache, which helped it actually a pretty good amount, though I don't think quite enough (I'm talking games, FPU intensive apps) to keep up with the CeleronA, which also maxed out at about 600mhz was the limit to that line, until they moved to .18 micron etc, compared to the K6-x line, which maxed out eventually at about 550mhz (those speeds weren't released till much later, when it didn't really count as much, becuase the Athlon was the main star for AMD).
the K6-2+ is a K6-2, with 128 k L2 cache on die, and the Athlons extra 3DNow! instructions included, as well as Powernow!, and made with the .18 micron die process (no copper as far as I know)
the K6-3+ is a K6-3, with the extra 3DNow! instructions that originally debuted on the Athlon, and PowerNow! as well, and is made with the .18 micron die process.
these CPU's seem to max out at about 600mhz, but do not produce as much heat as their predecessors. the reason they do not go as high as a .18 micron Celeron, is simply becuase they aren't designed to go that fast. problems occur with data when the K6-x line goes any faster.