Barracuda XT, Barracuda, Barracuda Green, DiamondMax and other desktop SATA and IDE drives

[babeliak]

 I have tested and compared 3 brand new sealed/unused (after unpacking I did full format in Win7x64) identical hard drives of the type ST2000DM001 (2TB Barracuda, 7200rpm, 2 platters @ 500GB/platter data density):

2 drives with MOD 3/2012 have performed wery well (I still have them, not using them as I need 3 drives for my RAID5),

1 drive has performed poorly, returned and replaced -->

1 replacement drive MOD 4/2012 is performing 25% worse (returned, awaiting replacement).

THESE ARE ALL IDENTICAL TYPE/SERIES = ST2000DM001, only redesigned slightly from 4/2012 onwards.

I talk about variation in between 2TB Barracudas!

Resume:
 

Newly redesigned drives from ST2000DM001 family with MOD from 4/2012 onwards are performing POOR - up to 25-30% WORSE !!!

Please see my test results, I compared 3 identical drives, I just returned the replacement drive too !!!

If I do not get a good performing drive next, I will return ALL THREE of them and give WD a try.

I bought this series drives based on user review here: overclockers.co.uk - user test

Now I make my unexpected bitter experience on my own.

Please new users, list your results with SN or DOM (Date Of Manufacture).

[fzabkar]

The difference is in the serial numbers, specifically the first three characters.

This is my understanding of their meaning (I could be wrong):

W = manufacturing location
1 = platter type/density (1 = 1TB/platter, 2 = 500GB/platter)
E = 4 heads (F = 6 heads, D = 2 heads)

Your first two drives have 4 heads whereas the slower drive has 6 heads. Even though they all appear to use the same platter type, the latter appears to have been low-level formatted at a reduced density.

If we imagine a smooth curve of best fit through all the benchmark data, then the faster drives would have a maximum sustained data transfer rate of about 210MB/s. The slower drive has a noisy curve, but let's say that it's transfer rate is around 175MB/s. The ratio of the transfer rates between the innermost and outermost zones for each drive is about 2:1, so none appear to have been short-stroked.

Therefore, I would apply the following rule of thumb:

(data density A)/(data density B) = [(data rate A)/(data rate B)]^2

Now (175 / 210)^2 = 0.694, so the slower drive appears to have a data density of ~700GB/platter. This suggests that it has 3 platters and 6 heads.

Here is the update matrix for the Grenada CC4C/4D/9D firmware update:
http://www.users.on.net/~fzabkar/HDD/GRCC4CD9.TXT

Notice that the ST2000DM001-9YN164 model has three possible physical configurations -- 4, 5, or 6 heads. Seagate's old model numbering system indicated the number of platters in the second-to-last numeric digit, but the new system obscures all these physical details.

[babeliak]

@fzabkar

Thank you for the info!

 I kindly asked the reseller for the WxExxxxx Series replacement drive and they delivered!

Now I have all three HDDs from WxExxxxx Series and NO PERFORMANCE ISSUES

I'm glad  you helped me solve my problem.

There should be NOT this big design and performance differences within ONE series of hard drive 

[jcvjcvjcvjcv]

@fzabkar;

Thank you very much! I knew of the existence of different versions of the ST2000DM001 and with your input on the serial number I could verify what I was buying. I just got a W1E ST2000DM001 and connected to a SATA 600 port on the P67 it gives me these scores in HD-Tune:

It's enterily empty and not an OS drive.

Minimum: 99.5 MB/s

Maximum: 208.8 MB/s

Average:163.0 MB/s

Access Time: 12.2 ms

Burst Rate: 221.6 MB/s

CPU: -1.0%

Connected to a SATA 300 port on the P67 gives just a bit lower score:

Minimum: 99.5 MB/s

Maximum: 197.7. MB/s

Average:161.9 MB/s

Access Time: 12.2 ms

Burst Rate: 154.1 MB/s

CPU: -1.0%

The main difference is in the Burst rate and the score in the very beginning of the drive.

Attached picture is with the drive connected to a SATA 600 port on the P67

!??!? No .PNG's allowed? Crazy..

[max789a]

> Notice that the ST2000DM001-9YN164 model has three possible physical configurations -- 4, 5, or 6 heads. Seagate's old model numbering system indicated the number of platters in the second-to-last numeric digit, but the new system obscures all these physical details.

>

Does this mean that a new ST2000DM001 could have two or three platters?

Thanks.

[jcvjcvjcvjcv]

Yes. That also explains the difference between the HD-Tune scores of 130MB/s and 160 MB/s

Also see:

http://forums.seagate.com/stx/attachments/stx/ata_drives/28488/1/2012-05-24_16-44-29_762.jpg

The drive on the left is the slower one. The recess in the top right corner is not as deep as the drives on the right.

[fzabkar]

Thanks for the photos.

Here is what I've been able to discern about the serial number:

Serial number prefix / Site Code / Place of Manufacture

S - SU - Suzhou China
W - WU - Wuxi China
Z - TK - Korat Thailand

In the following thread, a 2TB drive with a S/N of W2Fxxxxx actually had 5 heads. Therefore the third character appears to encode the number of platters rather than the number of heads.

http://forums.seagate.com/t5/Barracuda-XT-Barracuda-Barracuda/Force-firmware-update-on-ST2000DM001/t...

So F = 3 platters with either 5 or 6 heads, while E = 2 platters.

[fzabkar]

Correction, a site code of "TK" probably refers to the Teparuk plant in Thailand. That said, some URLs state that the Korat plant assembles the drives, while Teparuk manufactures the head stack components.

[Fernando]

There is NO SUCH THING as "CPU -1%" !

Is ALWAYS + never negative!

[Fernando]

Tested on USB3:

Date Code: 12415 (2012/04/13)  Site: WU