Different Conclusions Drawn by HDS and HDT

[rdpbug]

To start briefly, my question is about the different conclusions drawn by two hard disk diagnostic software, HD Sentinel (HDS) and HD Tune (HDT).

Here is a real case I got encountered recently. Based on the same S.M.A.R.T data from a FUJITSU MHZ2160BH G2 hard disk, HDS and HDT gave 2 different reports. While HDS said that the disk is still in perfect status, HDT gave a warning signal. The raw data of the two S.M.A.R.T diagnostic reports are attached for further information.

Report genertated by HDT:

Code: Select all

ID                               Current  Worst    Threshold  Data           Status   
(01) Raw Read Error Rate         100      100      46       211145         ok       
(02) Throughput Performance      100      100      30       29622272       ok       
(03) Spin Up Time                100      100      25       0              ok       
(04) Start/Stop Count            99       99       0        2297           ok       
(05) Reallocated Sector Count    100      100      24       8589934592000  warning  
(07) Seek Error Rate             100      100      47       2748           ok       
(08) Seek Time Performance       100      100      19       0              ok       
(09) Power On Hours Count        93       93       0        3904           ok       
(0A) Spin Retry Count            100      100      20       0              ok       
(0B) Calibration Retry Count     253      253      0        0              ok       
(0C) Power Cycle Count           100      100      0        2125           ok       
(B6) Unknown Attribute           100      100      0        0              ok       
(B8) End To End Error Detection  253      253      97       0              ok       
(B9) Unknown Attribute           253      253      0        0              ok       
(BA) Unknown Attribute           253      253      0        0              ok       
(BB) Uncorrectable Error Count   100      100      0        0              ok       
(BC) Command Timeout             100      100      0        0              ok       
(BD) Unknown Attribute           253      100      0        0              ok       
(BE) Airflow Temperature         63       45       0        622198821      ok       
(BF) G-sense Error Rate          253      99       0        16580611       ok       
(C0) Unsafe Shutdown Count       100      100      0        4653127        ok       
(C1) Load Cycle Count            98       98       0        53497          ok       
(C3) Hardware ECC Recovered      100      100      0        580            ok       
(C4) Reallocated Event Count     100      100      0        439943168      warning 
(C5) Current Pending Sector      100      100      0        0              ok       
(C6) Offline Uncorrectable       100      100      0        0              ok       
(C7) Interface CRC Error Count   200      253      0        0              ok       
(C8) Write Error Rate            100      100      60       31347          ok       
(CB) Run Out Cancel              100      100      0        1529006522271  ok       
(F0) Head Flying Hours           200      200      0        0              ok       

Health Status         : warning

Report genertated by HDS (S.M.A.R.T part):

Code: Select all

    S.M.A.R.T.
   ------------
No.  Attribute                Thre.. Value  Worst  Data                Status                   Flags                                                  
1    Raw Read Error Rate      46     100    100    0000000277B1        OK                       Error-Rate, Performance, Statistical, Critical         
2    Throughput Performance   30     100    100    000001C40000        OK                       Performance, Critical                                  
3    Spin Up Time             25     100    100    000000000000        OK                       Statistical, Critical                                  
4    Start/Stop Count         0      99     99     0000000008F9        OK (Always passing)      Self Preserving, Event Count, Statistical              
5    Reallocated Sectors Co.. 24     100    100    07D000000000        OK                       Self Preserving, Event Count, Statistical, Critical    
7    Seek Error Rate          47     100    100    000000000851        OK                       Error-Rate, Performance, Statistical, Critical         
8    Seek Time Performance    19     100    100    000000000000        OK                       Performance, Critical                                  
9    Power On Time Count      0      93     93     000000000F40        OK (Always passing)      Self Preserving, Event Count, Statistical              
10   Spin Retry Count         20     100    100    000000000000        OK                       Event Count, Statistical, Critical                     
11   Drive Calibration Retr.. 0      253    253    000000000000        OK (Always passing)      Self Preserving, Event Count, Statistical              
12   Drive Power Cycle Count  0      100    100    00000000084D        OK (Always passing)      Self Preserving, Event Count, Statistical              
182  Vendor-specific          0      100    100    000000000000        OK (Always passing)      Self Preserving, Event Count, Statistical              
184  End-to-End Error Count   97     253    253    000000000000        OK                       Self Preserving, Event Count, Statistical, Critical    
185  Vendor-specific          0      253    253    000000000000        OK (Always passing)      Event Count                                            
186  Vendor-specific          0      253    253    000000000000        OK (Always passing)      Self Preserving, Event Count, Statistical              
187  Reported Uncorrectable.. 0      100    100    000000000000        OK (Always passing)      Self Preserving, Event Count, Statistical              
188  Command Timeout          0      100    100    000000000000        OK (Always passing)      Self Preserving, Event Count, Statistical              
189  High Fly Writes          0      253    100    000000000000        OK (Always passing)      Self Preserving, Event Count, Error-Rate, Statistical  
190  Airflow Temperature      0      64     45     000025160024        OK (Always passing)      Self Preserving, Statistical                           
191  G-Sense Error Rate       0      253    99     000000FD0003        OK (Always passing)      Self Preserving, Event Count, Statistical              
192  Power off Retract Cycl.. 0      100    100    000000470047        OK (Always passing)      Self Preserving, Event Count, Statistical              
193  Load/Unload Cycle Count  0      98     98     00000000D0F8        OK (Always passing)      Self Preserving, Event Count, Statistical              
195  Hardware ECC Recovered   0      100    100    00000000018F        OK (Always passing)      Event Count, Error-Rate, Statistical                   
196  Reallocation Event Count 0      100    100    00001A390000        OK (Always passing)      Self Preserving, Event Count, Statistical             
197  Current Pending Sector.. 0      100    100    000000000000        OK (Always passing)      Event Count, Statistical                               
198  Off-Line Uncorrectable.. 0      100    100    000000000000        OK (Always passing)      Event Count                                            
199  Ultra ATA CRC Error Co.. 0      200    253    000000000000        OK (Always passing)      Self Preserving, Event Count, Error-Rate, Performanc.. 
200  Write Error Rate         60     100    100    00000000782E        OK                       Error-Rate, Performance, Statistical, Critical         
203  Run Out Cancel           0      100    100    0164002BFF87        OK (Always passing)      Statistical                                            
240  Head Flying Hours        0      200    200    000000000000        OK (Always passing)      Self Preserving, Event Count, Error-Rate, Performanc.. 

According to the results listed above, the main diversity lies in item 0x05 and 0xC4. Especially in terms of item 0x05, as I noticed, the raw data is 8589934592000 (DEC) shown in HDT which perfect equals 0x07D000000000 (Hex) shown in HDS. But it is interesting that it became 0 when I choose to show that value in decimal form in HDS. After thoroughly reading the FAQ and the introduction about the S.M.A.R.T on HDS website, I still have no idea whether item 0x05 is of special use on this hard disk or all the FUJUTSU disks because 0x07D000000000 is really extremely large compared with a normal case.

I'm now bewildered by these two reports. Shall I trust HDS and continue to use this hard disk just at ease, or take the warning signal raised by HDT into consideration and make a fully backup while preparing to find a substitution disk?

[hdsentinel]

Thanks for your message and question.

It is very common that different software give different results about a particular hard disk after inspecting the S.M.A.R.T. values of the drive.
This is caused by that most software simply can't understand / translate the attributes correctly.
I mean that most software simply translates the hexadecimal field as-is, showing an extremely large number (8589934592000 in this case) and issue a warning.
This is completely wrong approach of course.

Hard Disk Sentinel knows correctly that some attributes may contain different kind of data in the upper / middle / lower word of different S.M.A.R.T. attributes and which part(s) are related to actual error counters. This is the reason why
- Hard Disk Sentinel displays the attribute as hexadecimal value by default. This way advanced eyes quickly see that the attribute value is 07D000000000 - and the "rightmost" part (the middle and lower words, the center and rightmost 4 digits of the hexadecimal value) are all zeroes. These are the actual error counters, which are now zero.
- Hard Disk Sentinel displays the graph on the bottom by checking the "important" parts of the attributes and showing that the error counter is actually zero.
- Hard Disk Sentinel offers options to examine how the different parts (upper / middle / lower) of the selected attribute may change (see the "Attribute graph" option at the bottom right area on the S.M.A.R.T. page, next to the graph showing the change of the attributes)
- Hard Disk Sentinel has the proper text description and health % displayed to help your work (so it is not required to check the attribute values manually) - and does not display such invalid values to prevent confusion
- Hard Disk Sentinel offers various tests to verify the situation, reveal and fix any possible problems. These can (should) be used upon any situation where you are not sure that the reported value is accurate.

Personally I'd perform intensive testing by Hard Disk Sentinel (by the recommended methods, see the following links) - this would confirm if the situation is really perfect - or if there is any (even minor) problem with the hard disk.

So I'd recommend to check:

http://www.hdsentinel.com/faq.php#difference ( Why other software displays different health? Which is correct? )
http://www.hdsentinel.com/faq.php#tests ( Hard disk health is low or recently changed or I just installed a new (used) hard disk. How can I perform a deep analysis? )

as these would help to verify which information is more accurate

[rdpbug]

Thanks for the detailed clarification. I did notice that some S.M.A.R.T items are segmented meaningful instead of barely hexadecimal values as indicated in HDS S.M.A.R.T webpage, but I did not expect my hard disk just fall into this rare case. Last but not least, thanks to the professional introductions as well as suggestions on your website. Before reading your articles, I even used to take the S.M.A.R.T threshold as an alarm top-line for an uprising trend, which is in fact totally on the contrary.

[hdsentinel]

Thanks, I'm glad if I could help

To be honest, almost all hard disk have one or more such segmented S.M.A.R.T. attributes where the upper / middle / lower word indicates different kind of information. For example, look the "Airflow temperature" attribute as well: the different parts indicate different kind of measured temperature. So displaying these together as a simple large value (622198821) provided by other tool is completely wrong.
This is why it is really important to verify how the attributes on different models work - as without this knowledge the tools may provide incorrect/inaccurate results.

Yes, as you wrote originally the Threshold/Value comparison used to verify hard disk status and check if there are any problems. But this method confirmed that can't be used to report the real status and predict possible failures.
This is why the development of Hard Disk Sentinel started many years ago - and since that, more and more tools try to use the same method (to verify the Raw Data counters and report problems based on those) with more or less success...

http://www.hdsentinel.com/smart

page shows further information about this (maybe you already read that )