Run a benchmark, and tell us what you see.
http://www.hdtune.com/files/hdtune_255.exe
My HDTune results are in the left hand column. Usually, a rotating hard
drive, has a "declining" curve, as the diameter of the concentric tracks
changes as you get closer to the hub.
http://img829.imageshack.us/img829/842/500gb3500418ascomposite.gif
You can also use the "Health" tab, and look at the SMART statistics.
If the drive isn't all that healthy (failing), you might see an indicator
in there. If you see a "yellow" entry, that's OK. I have a couple
bogus yellow entries, and they're nothing. Since there are relatively
poor standards for SMART, sometimes the stats are misinterpreted.
The HDTune benchmark, doesn't check for alignment issues. The
partitions on a disk, can be aligned on multiple of 63 boundaries.
Or offset by multiples of 1 megabyte (Windows 7 method). If your
drive is new enough, it can be 4KB internal sectors, with "512e"
sector emulation for compatibility. And if a data partition is
misaligned, I suppose that could reduce performance.
The "64-bit" is a red herring, and has nothing to do with this problem.
Via DMA, the hardware takes care of copying the disk data into memory,
so when working well (non-PIO mode), the processor really isn't involved
in the transfer phase. Once in memory, the code will be written properly
by the driver or OS designers, for the most efficient transfer modes
if needed (cache-line aligned block transfers etc).
The high level copying code can suck. It sucked when Vista came out,
and that code was completely re-written at that point (new for Vista).
Since the code wasn't properly optimized before release (rush job),
it took a whole 'nother OS to attempt to fix it. So rather than software,
you could blame some aspects of that re-write for the issue. But
the issues weren't 32 bit versus 64 bit or anything.
This blog entry, is an example of a guy looking at why the
Vista copy code didn't work very well.
http://blogs.technet.com/b/markrussinovich/archive/2008/02/04/2826167.aspx
Paul