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

[oem7110]

I have two identical HDD, which is purchased around the same month, when I check health status, they both work fine.

referring to following linked image, the speed performance on the top HDD is very slow.
 
http://i1093.photobucket.com/albums/i438/junk000/HDD.jpg

Does anyone have any suggestions on what possible wrong it could be?
Any help will be greatly appreciated

Detailed discuss here, but the issue is not solved yet.

http://answers.microsoft.com/en-us/windows/forum/windows_xp-performance/what-wrong-is-my-hdds-readin...

[fzabkar]

It's not a PIO problem as was suggested in that other thread. If it were, then the transfer rate would flatline at a very low speed, probably below 5MB/s.

Clearly the slower drive (attached to the Marvell PCIe SATA controller) is being throttled by the controller. That's why there is a plateau of about 170MB/s at the outer zones. I don't know if it is significant, but the maximum transfer rate allowed by UDMA 7 is up to 167MB/sec. You may want to check HD Tune's Info tab. Otherwise check your system configuration or BIOS for anything that may be restricting the throughput of the PCIe bus. Notice that the burst rate is also limited to 165MB/s.

The big dips in performance probably indicate that the drives are stumbling on bad sectors, requiring excessive retries, but it could also be due to interference from background Windows tasks.

I would check the Health tabs. Look for reallocated, pending, or uncorrectable sectors.

[oem7110]

Please see following linked image for detailed description

Do you have any suggestions?

Thanks you very much for any suggestions

http://i1093.photobucket.com/albums/i438/junk000/HDD6.jpg

http://i1093.photobucket.com/albums/i438/junk000/HDD7.jpg

[fzabkar]

I can't see anything in your screenshots that would explain your drive's performance plateau. However, I don't believe that it is your drive that is the problem. IMO it's the Marvell controller. I expect that somewhere there may be a setting that limits the bus speed, but I don't know where to look. Sorry.

BTW, here is what PIO mode would look like:
http://img440.imageshack.us/img440/229/28dicembre20102329.png
http://www.neowin.net/forum/uploads/monthly_08_2010/post-25842-12818716727339.jpg

[oem7110]

There are what I find the major difference between slow and fast one.

Top (Slower)

Bottom (Faster)

Top (C5) Current Pending Sector is 1 for the worst
Bottom (C5) Current Pending Sector is 100 for the worst

Top (C6) Offline Uncorrectable is 1 for the worst
Bottom (C6) Offline Uncorrectable is 100 for the worst

What do following items indicate in term of performance?

Does anyone have any suggestions?

Thanks everyone very much for any suggestions

[fzabkar]

Sorry, I missed those numbers. I was looking at the raw values which are still zero. ISTM that the drive developed numerous bad sectors and set them aside pending reallocation. When the OS subsequently rewrote those sectors, the drive would have retested them. If they tested OK, then they would have been returned to service, otherwise they would have been replaced with spares. It appears that they all retested OK.

That said, since the normalised values of the two attributes dropped to their threshold levels, this would suggest that at one time there was a huge number of bad sectors. According to SMART attribute C2, the drive's maximum recorded temperature was 43C, so I can't imagine that this was responsible for the errors. Perhaps one of the heads slapped the surface and picked up some debris which degraded its read signal, and maybe this debris has since been dislodged???