ST31000528AS Seek Time\Lag

[Kyle6924]

Hello everyone! I've been having 1-3 second freezes while playing games, or listening to music, so I decided to start look into it. I ran a seek test on my drive, and it came back with an average of 16.41, but with maximum of 2365.73. So theres definately something wrong, but from this point I'm not really sure what. SMART is saying the drive is fine. Ive tried a new sata cable in a new port. I do not have another drive or computer that I can swap with to test. RAM and CPU tests come back fine, so I'm assuming its between the HDD\controller. Anyone have any insight for me?

Attachments

[Vulcan]

SMART is saying the drive is fine.

There are several possible ways to try to confirm your suspicion that the drive has a problem (and not some other part of the system).

One approach would be to download & run MHDD Scan (this requires the SATA controller BIOS setting to be IDE / compatibility / legacy or whatever your BIOS calls it) and look for unexpected pauses (and gather the full SMART data).

Another approach would be to start by looking at the full SMART data reported by HDD Sentinel, which you are already using. What is the full SMART data which it is showing? That's probably easier for you Can you attach that full SMART data in a screenshot?

Edited to add: Have you got a backup of whatever data you need from this drive?

[Kyle6924]

I do have a backup, so thats no problem. Heres the SMART data.

Attachments

[Kyle6924]

MHDD scan came back fine. I've run memtest 86 extended overnight, as well as windows memory diagnostics, both come back fine. Ive run prime 95 for a few hours with no issues, as well as a cpu burn in stability test. I'm fairly comfortable saying its not the ram, but do I need to take any additional steps to test the CPU?

[Vulcan]

Personally I don't agree that it's as simple as "If MHDD scans all sectors OK, then the drive is fine".

Can you take a photo of the MHDD screen (it doesn't need to be a multi-megabyte jpg, just good enough that we can read the text) after it has finished doing it's scan of this drive, and attach that to your reply?

In my experience, I've never seen the specific symptom which you describe (occasional slow I/O, but no I/O failures) caused by RAM or CPU.

[Kyle6924]

Yea, I'll do that in a bit. I don't think it's CPU or RAM, I'm leaning heavily towards the sata controller. It's just a little budget sandy bridge board.

[Kyle6924]

So its scanning right now, and under where it posts all the delays and such, its saying Warning: 234071385. What the heck is that?

[Kyle6924]

Alright, I ran scan on loop last night, heres the photo.

Attachments

[Kyle6924]

Does this give anyone a lead on what may be wrong?

[Vulcan]

Alright, I ran scan on loop last night

I appreciate that you're trying to provide info, but I didn't say to run it in a loop, for good reason Unfortunately from that screenshot, we can only see the most recent 25.1% of the result from MHDD.

heres the photo.

That shows that within the first 25.1% of the disk, there were 6 blocks (each of 255 sectors) with 50-149mS verify times, and 2 blocks with over 500mS verify times - which is obviously bad. Although I can think of one highly unlikely electronics issue which could cause this, it is much more likely that your drive is struggling to read a few sectors. I've seen this behaviour on those disks before - most sectors are normal & quick to read; a few are slower to read; and a very few are very slow (due to internal retries). And before you ask, no, this isn't necessarily visible in the SMART data, it all depends how far the drive's behaviour has degraded - been there, done that, got the T-shirt.

If you want more confirmation that it's a drive problem, then you need to check whether the slow areas are the same each time, for multiple single passes (if it's the same areas which are slow on multiple single passes, then it's the drive which is the cause). MHDD does have a way of making a log file during a scan, but I've only used that once, and you need to have MHDD on a writable medium with lots of space. Personally I wouldn't try. Instead I know HD Tach has a test of read throughput (i.e. sequential, not random, reads - sequential is what you need to use). If HDD Sentinel also has a sequential read test, then great, you can use that. In each case, you need to get a graph of the read throughput, print or save that graph from one full read of the disk - then do the same thing again, and compare the graphs. Do the dips in throughput line up on both graphs?

Make sense?