[lug] my "no-data-movement" file-copy attempt

[Doug Pintar]

Zan,
Thanks for your reply.  I really am an old disk hacker and started writing 
drivers for them way back in the early 1980s using Zilog Z-80 processors 
under a multi-user timesharing system (no kidding, it easily supported 8 
9600-baud serial terminals on an 8-bit microcomputer!) called "Oasis".  The 
original ST-506 interface, for which Adaptec (I think) made a "generic" 
controller that we were able to glue into an S-100 bus card.  Oasis, BTW, 
was seriously faster and feature-rich compared to the IBM PCs when they 
first came out.  Not as much address space per user process, but other than 
that the original MS-DOS/BIOS combo was quite the disappointment.  We 
basically used it as a program loader for our home-brew system and then 
kicked it out of the way, and I had to write what was basically a RAID 0 
driver (again using a lashed-up controller as the hard-drive-supporting 
PC-AT hadn't come out yet) for that one as well.  Anyway, I guess I let the 
similarity of the numbers for the 500 GB drive lead me off into lalaland. 
I'd sorta figured that the 250 GB drive had perhaps one less platter than 
the bigger one, rather than having bits that are almost "twice as wide" (or 
"long", track-wise) and take correspondingly longer to move around.  Your 
answer makes good sense, and I appreciate your taking the time to give me a 
concise, meaningful reply.  Just as an aside, both drives claim the full 
SATA 3Gb/sec transfer rate to/from their caches, but I suspect they lie.
Appreciatively,
Doug Pintar

----- Original Message ----- >
> Internal transfer rates are for just the drive electronics. It's usually
> the rate it can handle reading and writing from the cache on the drive.
>
> So the internal rate has nothing much to do with the rate that the drive
> can read from the platters.
>
> Beyond that, I am not surprised that a 250 GB drive is half as fast as
> the 500 GB drive. One hard drive performance parameter is the density of
> data on a track. A drive with double the data density will have double
> the transfer rate, if the spin rate and track spacing are the same.
>