Samsung 2TB HD204UI - 2 identical HDs, but work differently.

[dfr]

There are two Samsung 2TB HD204UI bought together (their S/N is different just by one digit at the end, so they had to be next to each other on the assembly line), but they behave very differently in tests.
Should I RMA the one with the big comb or saw like "teeth", worse random access time, and too many 500 ms surface access times?
Why exactly is this HDD behaving like that, what is the problem?

HDD 1:

HDD 2:

[unknown]

I think that you have to RMA both.

[dfr]

I think that you have to RMA both.

Why?

[unknown]

Both have lot of delayed sectors

[Vulcan]

@dfr:
I agree with unknown20010 - both sets of data you've provided have concerns...

Just to make sure that these are showing problems with the drives, and not problems caused by other parts of the system (although I think that's unlikely), it would be interesting if:

a) do the same tests on other drives show no large "dips" in throughput? Repeat that test several times to get confidence in the rest of the system, power, SATA cables etc. with other drives.
b) do the dips in throughput during testing on those two drives, always occur at the same part of the disk (i.e. the same horizontal position on the HDTach graphs)? (In my experience, the dips may not always be so "deep", depending on difficulty of reading that area, but the same LBA location is what is important). Again, repeat this a few times to be sure of whatever the answer is.
c) what does the full raw SMART data show for each of these drives? (Of special interest would be the number of reallocated & pending reallocation sectors.)

Those would be the additional tests that I would do & data to collect, if I was in this situation.

[dfr]

Both have lot of delayed sectors

Can you please check your HDDs for comparison?

[dfr]

@dfr:
I agree with unknown20010 - both sets of data you've provided have concerns...
[...]
c) what does the full raw SMART data show for each of these drives? (Of special interest would be the number of reallocated & pending reallocation sectors.)

Those would be the additional tests that I would do & data to collect, if I was in this situation.

At the moment I present SMART data using SpeedFan online in-depth analysis. Could you, please, for comparison, test with the same tool (SpeedFan online in-depth analysis) your HDDs and show the results in this thread? Many thanks.
For comparison I also tested an old Samsung (this one has several bad sectors) and WDC Caviar. The old Caviar looks better than the new Samsungs. Since they are new, they should have 100% health, and not 90%. Has anybody got the same Samsung model (HD204UI)? Could you please test it and show the results?

[Vulcan]

@dfr,

Unfortunately the "SpeedFan" data you have given, does not include the full raw SMART data from the disks that I explained before, which prevents a detailed review.

Examples of software which can give that raw SMART data include HDDScan (which you already have), and smartctl from the smartmontools suite. There are also others...

Anyway, I see that instead of performing my suggested tests, you are now requesting data from other members. Therefore I will leave you to carry on with your investigation, in the way that you have now decided. Good luck!

[dfr]

@dfr,

Unfortunately the "SpeedFan" data you have given, does not include the full raw SMART data from the disks that I explained before, which prevents a detailed review.

Actually the SpeedFan shows it, just the online in-depth analysis doesn't.

BTW, how do you read (analyse) raw data?

Examples of software which can give that raw SMART data include HDDScan (which you already have), and smartctl from the smartmontools suite. There are also others...

HD1:

HD2:

Anyway, I see that instead of performing my suggested tests, you are now requesting data from other members. Therefore I will leave you to carry on with your investigation, in the way that you have now decided. Good luck!

Come on! I said in my previous post "at the moment" meaning I cannot test them again as per your suggestion at the moment, I will do that later, but at the moment I checked out and can show the SMART data.
BTW, what's wrong with requesting for comparison data?

[Vulcan]

@dfr,
Oh dear, we seem to have misunderstood each other... Since board rules prevent off-topic discussions, I can't reply fully to your personal points and from your reply, I can see the potential for further communication / personality issues, so I'll withdraw from the thread after this post.

FYI, I did say that the "SpeedFan" data which you had (initially) given (i.e. "in-depth analysis") didn't include the raw SMART data - which was a correct statement. I don't use SpeedFan, so I had no idea that this data was available, yet you had just chosen not to include it - despite that being what I had requested. If you had provided that raw data initially, that would have saved me some typing, and helped to avoid misunderstanding

To reply to one of your points:

BTW, how do you read (analyse) raw data?

Using experience gained from around 2 decades of looking at similar data. There are public beginners articles on SMART and some of its interpretation e.g. on Wikipedia, to get you started.

I really did wish you good luck sincerely as in your reply, you seemed (to me) to have chosen your own path - which is fine with me, as it gives me more time for other things! Hopefully another board member who is better in-tune with your supply of data and your approach, will help you.

[dfr]

BTW, how do you read (analyse) raw data?

Using experience gained from around 2 decades of looking at similar data. There are public beginners articles on SMART and some of its interpretation e.g. on Wikipedia, to get you started.

But why is raw better IYO? Can you read (interpret) hexadecimal code?

I really did wish you good luck sincerely as in your reply, you seemed (to me) to have chosen your own path - which is fine with me, as it gives me more time for other things! Hopefully another board member who is better in-tune with your supply of data and your approach, will help you.

I provided you with the raw data, so what's your analysis of them?
From non raw data I can see that g-sensor has 2 points in the HD with the big "teeth".
Also since the "on-line indepth analysis" compares HDs to other user HDs of the same model, it seems that other people have these values different (they have green, what is orange in mine).