ST3250318AS High Raw Read Error Rate

[theabsinthehare]

I've had this hard drive for exactly one year, and I noticed recently that certain things hang when data needs to be pulled from the drive. Games will drop to extremely low framerates when loading new areas despite having more than enough ram and more than adequate video card. If anything is using the hard drive, like say I'm unzipping a file, other programs won't launch at all until the first app using the hard drive is done.

So, I opened HD Tune Pro, which I've had for quite a while but never use, and I noticed that the Raw Read Error Rate is *extremely* high; 125154961 to be exact. However, the status column says "Ok" for raw read error rate, and the Health Status of the drive says "Ok" as well. I ran an error scan and it showed no bad sectors, so I thought that number must be normal, so I checked my second drive, which is six years old, and it shows zero for Raw Read Error Rate.

There are also other high error rates in the list of stats for the newer drive, and oddly two negative stats for lifetime reads and writes. It doesn't make any strange noises, as people often say they hear when a drive is starting to fail. My power settings are set to spin the drive down after 30 minutes. The drive is connected via SATA. (My second drive is connected via PATA and slaved to an PATA CD/DVD drive if that could have anything to do with it)

This is the complete list of stats:

[cpp]
HD Tune Pro: ST3250318AS Health

Raw Read Error Rate Current: 117 Worst: 99 Threshold: 6 Data: 125498017

Spin Up Time Current: 97 Worst: 97 Threshold: 0 Data: 0

Start/Stop Count Current: 100 Worst: 100 Threshold: 20 Data: 161

Reallocated Sector Count Current: 100 Worst: 100 Threshold: 36 Data: 0

Seek Error Rate Current: 78 Worst: 60 Threshold: 30 Data: 79436688

Power On Hours Count Current: 91 Worst: 91 Threshold: 0 Data: 8688

Spin Retry Count Current: 100 Worst: 100 Threshold: 97 Data: 0

Power Cycle Count Current: 100 Worst: 100 Threshold: 20 Data: 77

End To End Error Detection Current: 100 Worst: 100 Threshold: 99 Data: 0

Reported Uncorrectable Errors Current: 100 Worst: 100 Threshold: 0 Data: 0

Command Timeout Current: 100 Worst: 99 Threshold: 0 Data: 65553

Airflow Temperature Current: 65 Worst: 51 Threshold: 45 Data: 655753251

Temperature Current: 35 Worst: 49 Threshold: 0 Data: 35

Hardware ECC Recovered Current: 64 Worst: 49 Threshold: 0 Data: 125498017

Current Pending Sector Current: 100 Worst: 100 Threshold: 0 Data: 0

Offline Uncorrectable Current: 100 Worst: 100 Threshold: 0 Data: 0

Ultra DMA CRC Error Count Current: 200 Worst: 200 Threshold: 0 Data: 0

Head Flying Hours Current: 100 Worst: 253 Threshold: 0 Data: 8887

LifeTime Writes from Host Current: 100 Worst: 253 Threshold: 0 Data: -368822297

LifeTime Reads from Host Current: 100 Worst: 253 Threshold: 0 Data: -825534820

Health Status : ok

[/cpp]



Is my drive defective?

 

[fzabkar]

There is nothing in your SMART data that should give you real cause for concern, except possibly the command timeouts.

Seagate has counterintuitive SMART attributes. Furthermore, HD Tune's author only reports the lower 32 bits of each attribute, not the full 48 bits.

The raw value (125498017) for the Read Error Rate attribute is actually a sector count, not an error count. This count rolls over to zero after 250000000. The normalised value of the attribute is logarithmic and is calculated according to some proprietary formula.

The raw value of the Seek Error Rate (79436688) stores the total number of lifetime seeks in the lower 32 bits, and the total number of seek errors in the uppermost 16 bits. In your case SMART has recorded no errors in 79436688 seeks.

The normalised value of SER is given by the formula ...

SER = -10 log (total seeks / total seek errors)

Therefore, even though intuitively it looks bad, your SER is a perfect score.

The Command Timeout attribute looks a lot better when viewed in hexadecimal.

You can use Google's calculator for smaller numbers:
http://www.google.com/search?q=65553+in+hex

65553 (decimal) = 0x000000010011

I believe the actual number of timeouts is stored in the lowest 16 bits, ie 0x11 (= 17 decimal).

The LifeTime Writes from Host and LifeTime Reads from Host attributes are incorrectly displayed as negative numbers but will be correctly displayed in hexadecimal.

The raw value of Airflow Temperature (655753251) also makes more sense in hex.

655753251 (dec) = 0x000027160023 (hex)

There are actually three temperatures in the above number -- 0x27 (=39C), 0x16 (22C), and 0x23 (35C). These probably represent the max/min/current temperatures for the current power cycle.

BTW, the normalised value of the Airflow Temperature is derived by subtracting the value of the Temperature attribute from 100.

HD Sentinel and HDDScan both display the full 48-bit values in hex.

BTW, I find it strange that Head Flying Hours (8887) is greater than Power On Hours (8688). I've seen this several times before on Seagate drives.