Maybe it had something to do with the dingo, lol.
What you're describing would definitely not go unnoticed if it were a unified phenemonon that was reproducible across all configurations with particular processors; ie; every E8500-based system had lag that was impossible to eliminate in general windows usage. You're correct in that many general users probably wouldn't notice the difference between .1s and .7s time to open notepad or calc for example, but you know that people like us supernerds on AT would notice that in a heartbeat. We could make videos, and the word would spread like wildfire. This would have particularly been pronounced at the beginning of the C2D era, when it was a shocking development to see Intel go from the back of the pack to a commanding lead overnight. Much like the Nvidia FX5xxx series debacle, and the ATI 2900 mis-step, die-hards from either camp would go to great lengths to seize upon any possible shred of evidence to back up their team, and if there were evidence, we'd definitely have seen it at the time.
Remember that the X2-5000+ was only available in Windsor and Brisbane forms, and all were 512k L2 cache units. One of the closer things I can think of that is a test of real-world performance in general windows/office is the old PC Worldbench, which times response times to perform simple tasks (open word, change font, save test file, etc). Check out this old AT comparo at the $180 price point waaaaaaaaay back in '07 :
http://www.anandtech.com/show/2276/6
Notice that the X2 6000+ (marginally better than the X2 5000+) is handily beaten despite a clock speed advantage by the old E6750, including the worldbench 6 test.
Now consider this, the E6750 was an old 2.66ghz CPU with 4mb of cache, running on old chipsets with older memory. The E8500 is a 3.16ghz CPU with 6mb of cache, and they typically were mated to much newer chipsets and better memory. If the old E6750 tested better in running those simple routines against a 6000+, how is a 5000+ supposed to be faster than an E8500. Does not compute, literally.
Again, and I can't say this clearly enough, I absolutely believe you saw/witnessed what you describe, as I've seen similar oddities myself in all kinds of setups. I can say with all confidence that the following is true :
(1)- Your X2 5000+ system is working perfectly.
(2)- The E8500 system is fubared somewhere. Not only is it lagging with a faster HDD, but the hitching in FO3 that you describe is absolutely an indication of some more serious problem lurking under the surface.
(3)- The E6800 Pentium Dual Core system in the big box store is virtually useless as a comparitive tool as the OEMs gum up PCs so badly out of the box that it's borderline criminal. I've seen brand-new PhII X6 and Core i7 boxes with 8GB or 12GB of ram and fast hard drives/etc running like dog crap because they have 90 or 100 processes running and so much crapware/demoware that it just brings Windows to a disgusting crawl. I'm a picky SOB, and my old extra-box E5200 (much slower than E6800) produced no appreciable lag in Windows Vista x64 or Windows 7, even at stock speeds, on a low-level G31-based mobo and generic DDR2, with a mediocre hard drive. I'm talking about the performance you probably see with your 5000+, literally by the time I'm lifting my finger from the second double-click, any simple task was already complete and waiting. And I played all of FO3 on that setup with no hitching
I'm not saying that I'm some all knowing genius or anything of the sort, but I slog through this crap 50+ hours a week, year after year, migrations, mass loadset rollouts, clean and wipe/reloads for new hires, and although some weeks I might be up to my ears in purely software/server/remote work, most weeks I have to do minor to serious hands-on work with 5-10 desktops or laptops. I'd have definitely noticed if there were something to this. About the only thing I have noticed that's relevant is the oddballs that seem out of place. Like having a P4 or Sempron run like a top, or running into a quad-core system with plenty of ram and no crapware running like a dog. Most of the time an answer can be found, but sometimes it's just not easy to find or worth an infinite investment in time to grind out the root cause.