Would you like to supply proof that backs your statement, because the
millions who do see better all round performance will be bemused by
what you say..
Well Mike, I would have liked to have done a test myself, as I just built
a new PC for my mother-in-law. (Would an AMD64 3600+ w/1Gig of RAM have
qualified ?)
Here's the problem(s) I saw with doing a test though.
The amount of time involved, for me, would be enormous.
First, I'd have to write a test plan.
Second, I'd have to try to find all of the latest drivers for each piece
of hardware. Not a problem, but if these drivers were not MS 'certified',
which probably would have been the case, some would cry that the test
results don't count because they weren't certified drivers.
Third, what to use to measure performance. Since 'seems to' and 'I get
the impression' will absolutely not be recognized as valid, benchmarking
s/w would need to be used. So now you will have many people crying that
the benchmark software is just that, and that's not really how a PC
functions in real life.
Then, there would have to be 2 tests, one side by side 'out of the box'
test between XP and Vista, and then another after each OS gets tweaked as
best as can be.
Now, each and every tweak would need to be listed for each OS, in exact
detail. This would need to be done to keep people quiet.
The entire excercise would also need to be video-taped, un-edited, as to
prevent anyone form claiming that something else was done/modified which
affected (either) system which wasn't shown, which would make the tests
invalid.
I did consider it, but after realizing that most of the regulars here
would not believe any test results, no matter how well-documented and
methodical the tests are done.
So after all is said and done, no matter how much work would be put into
the testing and how valid the procedures really were, it would mean
nothing.
To take it even further, even if XP did come out on top, the results
would mean nothing anyway, since that would be attributed to 'well it's
only one machine', XP may be faster on that particular machine.
So, a big waste of time, unless a minimum of 10 PCs were involved with
vastly different hardware devices between them.