Prev: S.M.A.R.T. reallocated sector count failure
Next: SES-2 over I2C == multimaster I2C?
From: Olin K. McDaniel on 19 Oct 2006 16:05
I just bought a new Western Digital USB external hard drive, 250 GB in
size, called "My Book Essential" and am trying to use it to store some
large backup files created with Nova Back, vers. 8 - I think. The
files are roughly 4 GB in size with extensions of .nb7, thus the
confusion on the actual version at the moment. These files are all on
an internal IDE hard drive, and the computer has Windows 2000 on it,
and it is claimed to be good for USB 2.0, with 8 such sockets. The
CPU is an AMD 3100+.

My problem is - these files seem to be taking FAR too long to copy
over, for what USB 2.0 claims to be capable of. That claim is for 480
Mb/s, which I believe should be the equivalent of about 40 MB/s. For
whatever reason these 4 GB files are consistently taking 75 minutes
to copy over, which seems absurdly too long for USB 2.0. It's more
like USB 1.0 or 1.1.

Has anyone else out there got any experience in this area, with this
"external" or "Essential" drive? The particular model number of this
one is WD2500D032 or the Order No. is WDG1U2500N. I've already plowed
around W.D.'s web site, their knowledge base, etc. and still do not
have an answer. Any help will be appreciated.

Olin McDaniel
From: craigm on 19 Oct 2006 17:11
Olin K. McDaniel wrote:

> I just bought a new Western Digital USB external hard drive, 250 GB in
> size, called "My Book Essential" and am trying to use it to store some
> large backup files created with Nova Back, vers. 8 - I think. The
> files are roughly 4 GB in size with extensions of .nb7, thus the
> confusion on the actual version at the moment. These files are all on
> an internal IDE hard drive, and the computer has Windows 2000 on it,
> and it is claimed to be good for USB 2.0, with 8 such sockets. The
> CPU is an AMD 3100+.
>
> My problem is - these files seem to be taking FAR too long to copy
> over, for what USB 2.0 claims to be capable of. That claim is for 480
> Mb/s, which I believe should be the equivalent of about 40 MB/s. For
> whatever reason these 4 GB files are consistently taking 75 minutes
> to copy over, which seems absurdly too long for USB 2.0. It's more
> like USB 1.0 or 1.1.
>
> Has anyone else out there got any experience in this area, with this
> "external" or "Essential" drive? The particular model number of this
> one is WD2500D032 or the Order No. is WDG1U2500N. I've already plowed
> around W.D.'s web site, their knowledge base, etc. and still do not
> have an answer. Any help will be appreciated.
>
> Olin McDaniel


Are you running W2k with SP4? Do you actually have USB 2.0 drivers running?
From: Ed Light on 19 Oct 2006 23:56
According to HD Tach, my usb 2 enclosure with my Seagate Barracuda IV in it
communicates at 29 mb/s but the average serial read is 18. Out of the
enclosure the average read is in the 30's. I think that's pretty typical.
Can't say why.

You might download it and see what it says about yours.
--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
spam(a)uce.gov
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org



From: craigm on 20 Oct 2006 09:48
Ed Light wrote:

> According to HD Tach, my usb 2 enclosure with my Seagate Barracuda IV in
> it communicates at 29 mb/s but the average serial read is 18. Out of the
> enclosure the average read is in the 30's. I think that's pretty typical.
> Can't say why.
>
> You might download it and see what it says about yours.


Serial protocols have significant overhead. As you add an additional level
of complexity to the data transfer, the throughput goes down.
From: Folkert Rienstra on 20 Oct 2006 11:47
"craigm" <none(a)domain.invalid> wrote in message news:Sr4_g.7$Y97.153749(a)news.sisna.com
> Ed Light wrote:
>
> > According to HD Tach, my usb 2 enclosure with my Seagate Barracuda IV in
> > it communicates at 29 mb/s but the average serial read is 18. Out of the
> > enclosure the average read is in the 30's. I think that's pretty typical.
> > Can't say why.
> >
> > You might download it and see what it says about yours.
>

> Serial protocols have significant overhead.

All protocols have overhead, serial just has more.

> As you add an additional level of complexity to the data transfer,

You need more raw transfer speed (bus speed) to transfer the same
amount of user data.

> the throughput goes down.

Nope, that only applies to the transfer rate ceiling.
That defines a maximum possible user data throughput for the bus.
If the needed raw throughput is more than the serial bus can deliver.
Everything below that fits within the available bandwidth and is not limited.

That does explain(?) the 29MB/s userdata transfer rate vs the 60MB/s
(480Mb/s) serial bus clock but doesn't explain the low average of 18.
Based on the average 35MB/s or so his drive can do externally that gives
you roughly 45MB/s on outer zones, 22MB/s on inner.

That should give you around (29+22)/2 = 26MB/s average.
 |  Next  |  Last
Pages: 1 2 3
Prev: S.M.A.R.T. reallocated sector count failure
Next: SES-2 over I2C == multimaster I2C?