FWIW, here are Vista's performance benchmarks on a new system, before and after upgrading RAM. This might help you decide what to expect if you upgrade.
System: Semprom 3200, 1.8 GHz, shared video memory, Vista Home Basic.
Item Before After
RAM 512 MB 2 GB
533 MHz 800 MHz
Processor 3.8 3.8
RAM 2.0 4.9
Graphics 3.8 5.1
Game graph 2.3 3.0
Disk 5.4 5.4
Sorry, but Microsoft's so-called "performance score" is for all
practical purposes can be misleading.
Presently the highest score you can get is 5.9 which I get for both
memory and graphics with a little less for Processor, hard disks.
Overall my system gets a 5.5 which isn't a average of the five test
areas but is the lowest of the five according to Microsoft. Anyway you
look at it a pretty fast system.
Presently I'm overclocking my CPU, a Intel Dual Core 6400 rated at
2.13GHz, a modest amount. Being in a playful mood I rebooted and
entered BIOS and turned off all overclocking. If the "performance
testing" actually did something you would expect to see performance
drop like a rock with overlclocking turned off. It didn't budge at
all.
Next I rebooted after restoring my overclocking setting to what they
were then I brought up CPU-Z (nice tool everybody should download)
that shows how well your CPU and memory and key things like your front
side bus speed (FSB) are actually doing. According to CPU-Z my 2.13
GHz CPU is actually running at 2.5 GHz with overclocking in part
because the FSB is pumped up to 1,239 MHz. Clearly if Vista was doing
any real "testing" it should have been reflected in some difference
between running at just the CPU's rated speed and overclocked. It
would seem all the performance test does is read the rated value of
the CPU rather then test it like CPU-Z actually does under load. By
the way to really get a realistic report I was rendering a video, one
of the most intensive tasks you can ask any computer to do during the
testing. ;-)