HDD GURU FORUMS

Hi,

I bought this SSD around a year ago but never got around to using it. I did a clean install of Windows 8.1 and ran HDTunePro and found a few errors which I'm not so sure of:

Also, the benchmark for HDTunePro returned around 180mbps over SATA6, I'm sure this is cause for worry alone?

Any suggestions?

Check it again after about a week of use and see if the bad block count increases. Back up regularly.

I don't understand how a single Retired Block can cause the health of that attribute to drop from 100 to 1, unless the meaning of attribute 05 has changed for this model.

Doesnt look like the util has been updated in 14 months. I am hesitant to trust it on SSD seeing as they are touting improved SSD support back then.. leading me to believe it wasnt all that great before whatever improvements. I guess it is simply reading SMART, but I dont know how different SSD SMART is to Conventional drives.

I would be looking for a current util with good SSD suport

AFAICT, smartmontools has the best SMART support.

For consistency purposes, always check smart values with more then one tool. So many available out there...

Every tool should report the same SMART values, although some tools misinterpret the raw values. It's the names of the attributes that vary widely, and that's because the drive doesn't actually report the names, only the attribute IDs.

My concern with many of these tools is that their authors don't understand how the raw values are encoded, and this then leads them to make erroneous assessments of the drive's health. Some authors go one step further and impose their interpretations on the user without displaying the actual raw data.

@fzabkar
I agree, especially when it comes to flash based devices, therefore suggesting checking with several tools and comparing to the main one anybody chooses to stick with as the most trusted.

Also, when it comes to flash, the problem can increase in complexity when people alter these values and then attempt to sell them as used devices on eBay and the like.

Ideally, I would recommend to buy "established" brand drives from a manufacturer that makes available smart utilities developed by them directly. Not sure if any are available right now, but it should become more widely available in the near future as this topic on flash will become hotter and hotter as time passes on.

To OP

Post here a SMART report from CristalDiskInfo tool....The program support almost all SSD's on the market.

Its funny becuase it says 2.5TB of info has been read and 5.6TB has been written... That is ludacris lol becuase I have only just managed to use up 200GB of the drive. It is impossible to have written 5.6TB of data.

Crystal Screenshot:

It could be counting testing at the factory, also when you write data to flash devices, there is other operations such as error correcting code, page and block re-writes, optimisations like delayed write and wear levelling. So the data written is not exactly the same size as what is recorded as written.

It does seem a fairly large amount though

Short version:

AFAICT, between the two SMART reports, about 400GB of data have been written to the drive, and about 350GB of data have been read from the drive.


Long version:

AFAICT, the Current and Worst values of the Accumulated Runtime Bad Blocks attribute are raw counts rather than normalised health scores. Comparison with other SMART reports suggests that a drive begins life with zeros for both values.

Here are 3 examples:

http://www.oczforum.com/forum/showthrea ... (unplanned)
http://www.ocztechnologyforum.com/forum ... 13473.html
http://forum.notebookreview.com/solid-s ... -data.html

Also, a normalised value of 80 for the Available Over-Provisioned Block Count appears to be how the drive begins life, even though one would expect that a perfectly healthy drive might show 100 for this attribute. The raw numbers appear to reflect the number of OP sectors (or LBAs) rather than NAND blocks. Also, over-provisioning seems to be achieved simply by virtue of the difference between decimal gigabytes and binary gigabytes.

For example, the OP's 256GB Vector would have 16 x 16GiB NAND chips for a total capacity of 256GiB. The actual number of reported LBAs is not given in either screenshot. However, the 512.1 GB example below reports 1000215216 sectors.

The total number of sectors in a 512GiB array would be ...

512 x 1024 x 1024 x 1024 / 512 = 1073741824

The difference in the decimal and binary capacities is ...

1073741824 - 1000215216 = 73526608 sectors

... which roughly corresponds to the raw value of attribute 171 (73518416) in the example below.


OCZ-VECTOR 256.0 GB
AB Available Over-Provisioned Block Count 80 80 0 22C2D50 (= 36449616)

OCZ-VECTOR 512.1 GB 1000215216 sectors
AB _80 _80 __0 00000461CD50 (= 73518416) Available Over-Provisioned Block Count

OCZ-VERTEX450 128,035,676,160 bytes
171 Unknown_Attribute 080 080 000 - 17907024

OCZ-VECTOR 128GB
171 Available OP block count 0x0000 080 080 18365776

OCZ-VECTOR 128GB
171 Available OP block count 0x0000 080 080 18185552


As for the amount of data written and read, ISTM that attributes F1 and F2 count the amount of data transferred over the SATA interface, whereas attribute F9 records the number of actual NAND writes, including wear levelling, etc. Therefore F1 and F2 should accurately reflect the user's usage pattern.

In fact we could compare the CrystalDiskInfo and HD Tune reports and see whether the differences in the attribrute values correspond to the actual usage over the intervening 10 days.

The Power On Hours Count differs by 166 hours (= 328 - 162).

The Average Erase Count has increased from 31 to 37.

The increase in Host Writes = 0x160F - 5250 = 397GB

The increase in Host Reads = 0xA06 - 2215 = 351GB

The increase in NAND Programming Count = 0x1A1AAEED - 388567443 = 49 388 890

CrystalDiskInfo believes that the Total Host Writes/Reads are reported in GB, and it appears that it assumes that each NAND Programming cycle addresses 16KB.

According to CrystalDiskInfo, 0x1A1AAEED NAND writes equates to 6682 GB.

So ...

6682 gigabytes / 0x1A1AAEED = 16382.3 bytes

Therefore the additional NAND writes between the two SMART reports amount to ...

49388890 x 16KiB = 753.6 GiB

That's a write amplification of approximately 2:1.

AFAICS, the difference in the Average Erase Count would suggest that the entire NAND array has been erased 6 times, which amounts to 1.5TB (= 6 x 256GB). That's twice the increase in NAND writes. Does that make sense, or should the two results be roughly equal, in which case the write amplification would be 4:1, and each NAND programming cycle would address 32KB?

I'd love to reply but what would the point be?

The SSD seems to have failed all so randomly just a moment ago (or so I think anyway.)

The PC froze and hanged for a good 5 minutes before I decided to CTRL ALT DEL and all that resulted was a black screen, so I went ahead and restarted the PC from the restart button.

All this caused was the boot screen to hang and then a message to appear "CSM not recognized. Please go into the boot settings and adjust."

When I did, no HDD was listed. Nothing.... So it looks like it really was looking to go up in flames and it just did. It doesn't even appear in the main menu as a drive.

So I guess this is it then?

EDIT: Now my PC completely refused to turn on. W...T...F.

I removed it and connected it to an external caddy and tried detecting it from my Linux Mint laptop but nothing appeared.

Does this mean my worst fears are confirmed? If it is, then this is 10x worse then loosing the files I did on my 3TB backup drive as everything on here was all recent and all very important and it will cost me big time, better yet... priceless.

All my work.... gone.

If the PC doesn't turn on, then the PSU may be faulty. Do the fans kick at all? If so, then I'd suspect a short on one or more of the devices.

Have you checked the PSU voltages with a multimeter? Is there a 5VSB (standby) voltage? With the PSU disconnected from the AC mains supply, and with all the devices connected to the PSU, what resistances do you measure between ground and each of the outputs, especially +5V and +12V?

fzabkar wrote:If the PC doesn't turn on, then the PSU may be faulty. Do the fans kick at all? If so, then I'd suspect a short on one or more of the devices.

Have you checked the PSU voltages with a multimeter? Is there a 5VSB (standby) voltage? With the PSU disconnected from the AC mains supply, and with all the devices connected to the PSU, what resistances do you measure between ground and each of the outputs, especially +5V and +12V?

Hi, I tried to edit the post but couldn't. The PC works fine now so ignore that statement, no idea why it was refusing to turn on.

I tried the SSD using an external caddy on my Linux laptop but it failed to recognize. I tried another drive just to make sure and it picked up so this drive is definitely faulty somehow.

I just checked OCZ's website and a firmware update was announced last May:

Improvements since version 1.03.1

[*]Fixed a corner case issue with firmware corruption after unexpected sudden power loss scenarios, which caused drives to no longer be recognized.
[*]Improved SATA timings, which addresses intermittent detection issues on some platforms.

The problem you describe is a very common one. The storage forums are full of similar incidents. I wouldn't touch OCZ's products with a barge pole. Sorry for your loss.

So its absolutely unrecoverable?

My main priority would be to recover the spreadsheet files, there's maybe 50mb's worth so its a very small recovery thats needed, but the data in them is very important.

It may be possible to perform Flash recovery depending on what chipset it is. I don't recall seeing you post that anywhere. If it is Sandforce controller, I don't think there is a solution for most, if not all of them. If it is intel or Indilinx like an 1100 or something then a possibility.

I could do a recovery if you want to post depending on the chipset.

Do not start fiddling around with it until you have a good ide what each operation you try on it does.


On 6th you started posting you were concerned about the drives behaviour. The very first thing you should have done, even before asking the initial question here was backup. The fact you already lost a 3TB should have re-inforced that. It is a really painful thing to lose files, and I hope that if nothing else, others will read this and instead of focussing on what does this number mean, why is my drive weird, what utils can I harass the drive with next - instead they backup everything so it really doesn't matter what the dodgy drive does. I think people forget that once a drive has any sort of issue, it may not be possible to predict the next escalation of the issue no matter how much information you have on it.
If there is any firmware bug, corruption or an issue with the drive then unless you know exactly what that is and can take that into account when looking at the drive info, stats , settings and parameters... then you cant know what it is going to do next.

lets hope that there is a chance of recovery if the chipset is a sane one.

Slight possibility it could be a pcb issue did you want to ship it to AU?

all the best

PentiumCore wrote:The PC works fine now so ignore that statement, no idea why it was refusing to turn on.

If the SSD has a shorted TVS diode, then that may explain why the PC wouldn't turn on when the SSD was installed inside the PC, and is now OK after moving the SSD to an external enclosure, assuming that was the sequence of events.

Do you have a multimeter, and can you upload detailed photos of both sides of the SSD's PCB?