Prev: external enclose 3.5" eSATA recommendations? (or not-recommendations)
Next: BEST BRANDS PORTAL
From: Rod Speed on 9 Sep 2009 20:45
Franc Zabkar wrote:
> On 7 Sep 2009 13:34:33 GMT, Arno <me(a)privacy.net> put finger to
> keyboard and composed:
>
>> Franc Zabkar <fzabkar(a)iinternode.on.net> wrote:
>>> Looking at Table 1 on page 10 of the following document, I see that
>>> the 1.5TB and 2.0TB drives both have 4 platters and 8 heads, and
>>> identical data densities.
>>
>>> Barracuda LP Series SATA Product Manual:
>>> http://www.seagate.com/staticfiles/support/disc/manuals/desktop/Barracuda%20LP/100564361b.pdf
>>
>>> In fact the only difference in the specs is the number of guaranteed
>>> sectors.
>>
>>> Why wouldn't Seagate have used 3 platters and 6 heads?
>>
>>> I considered the possibility that, for the same capacity, more
>>> platters means less cylinders, and therefore better average seek
>>> times, but this isn't reflected in the specs.
>>
>>> Surely this isn't a yield issue?
>>
>> Why not? The 1.5TB model may just be a way to not have to
>> scrap 2TB models that do not make the cut. Also, producing
>> 2TB drives and limiting them to 1.5TB may be cheaper than
>> having two assembly lines for 1.5TB and 2TB both. Would
>> not be the first time...
>>
>> Arno
>
> If Seagate is producing drives with 4 fully functional platters and
> then turning off 2 heads, why not take full advantage of the potential
> performance gains?

Because its a lot harder to do and that particular market is entirely driven by price.

> Reducing the number of cylinders instead of turning off an entire
> platter would result in better average access times and higher average
> throughput (because the outer cylinders pack more data per revolution
> than the inner ones).

And is a lot harder to do than just not using a couple of heads on the less than ideal platter.


From: Franc Zabkar on 17 Sep 2009 03:26
On 10 Sep 2009 00:15:31 GMT, Arno <me(a)privacy.net> put finger to
keyboard and composed:

>Franc Zabkar <fzabkar(a)iinternode.on.net> wrote:

>> If Seagate is producing drives with 4 fully functional platters and
>> then turning off 2 heads, why not take full advantage of the potential
>> performance gains?
>
>> Reducing the number of cylinders instead of turning off an entire
>> platter would result in better average access times and higher average
>> throughput (because the outer cylinders pack more data per revolution
>> than the inner ones).
>
>Because this would make the strategy obvious?
>And in addition, there is the yield question.

I was wondering how WD's Raptor drives were achieving 8ms access times
while their other models were getting only 14ms:
http://www.hdtune.com/testresults.html#Western_Digital

It seems that WD's 10K RPM drives had smaller 3.0" platters compared
with the usual 3.25":
http://forums.macrumors.com/archive/index.php/t-200923.html

My calculations (for Fujitsu drives) suggest that the usable data area
occupies a band of width 2.37 cm:
http://groups.google.com/group/microsoft.public.windowsxp.hardware/msg/b4eb3d937a08e2e3?dmode=source

Assuming the reduced platter size results in a reduction of 6mm in the
usable radius, then that equates to about 73%. That, and the reduced
rotational latency, explains the faster access times.

So, if WD got away with it, why couldn't Seagate? ;-)

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.
From: Arno on 17 Sep 2009 21:33
Franc Zabkar <fzabkar(a)iinternode.on.net> wrote:
> On 10 Sep 2009 00:15:31 GMT, Arno <me(a)privacy.net> put finger to
> keyboard and composed:

>>Franc Zabkar <fzabkar(a)iinternode.on.net> wrote:

>>> If Seagate is producing drives with 4 fully functional platters and
>>> then turning off 2 heads, why not take full advantage of the potential
>>> performance gains?
>>
>>> Reducing the number of cylinders instead of turning off an entire
>>> platter would result in better average access times and higher average
>>> throughput (because the outer cylinders pack more data per revolution
>>> than the inner ones).
>>
>>Because this would make the strategy obvious?
>>And in addition, there is the yield question.

> I was wondering how WD's Raptor drives were achieving 8ms access times
> while their other models were getting only 14ms:
> http://www.hdtune.com/testresults.html#Western_Digital

> It seems that WD's 10K RPM drives had smaller 3.0" platters compared
> with the usual 3.25":
> http://forums.macrumors.com/archive/index.php/t-200923.html

> My calculations (for Fujitsu drives) suggest that the usable data area
> occupies a band of width 2.37 cm:
> http://groups.google.com/group/microsoft.public.windowsxp.hardware/msg/b4eb3d937a08e2e3?dmode=source

> Assuming the reduced platter size results in a reduction of 6mm in the
> usable radius, then that equates to about 73%. That, and the reduced
> rotational latency, explains the faster access times.

> So, if WD got away with it, why couldn't Seagate? ;-)

WD is (or used to be) the only HDD company without SCSI drives.
Seagate does this in their SCSI models and you pay for it.

Arno

First  |  Prev  | 
Pages: 1 2
Prev: external enclose 3.5" eSATA recommendations? (or not-recommendations)
Next: BEST BRANDS PORTAL