Bad sectors on SSDs

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ao1

Member
Apr 15, 2012
122
0
0
From a NAND product data sheet.....

“This NAND Flash device is specified to have a minimum of 3,936 MLC blocks (NVB) out of 4,096 total available blocks. This means the devices may have blocks that are invalid when shipped from the factory. An invalid block is one that contains one or more bad bits. Additional bad blocks may develop with use. However, the total number of available blocks will not fall below NVB during the endurance life of the product.

Although NAND Flash memory devices may contain bad blocks, the devices can be used quite reliably in systems that provide bad-block management and error correction algorithms. This type of software environment ensures data integrity.

Internal circuitry isolates each block from other blocks, so the presence of a bad block does not affect the operation of the rest of the NAND Flash array.

The first block (physical block address 00h) for each CE# is guaranteed to be valid with ECC when shipped from the factory.

NAND Flash devices are shipped from the factory erased. The factory identifies invalid blocks before shipping by attempting to program the bad-block mark into every location in the first page of each invalid block. It may not be possible to program every location with the bad-block mark. However, the first spare are location in each bad block is guaranteed to contain the bad-block mark. This method is compliant with ONFI Factory Defect Mapping requirements.

System software should check bytes 4,096 to 4,319 on the first page of each block for a 00h value prior to performing any PROGRAM or ERASE operations on the NAND Flash device. A bad-block table can then be created, allowing system software to map around these areas. Factory testing is performed under worst-case conditions. Because blocks marked “bad” may be marginal, it may not be possible to recover this information if the block is erased.

Over time, some memory locations may fail to program or erase properly. In order to ensure that data is stored properly over the life of the NAND Flash device, the following precautions are required: • Check status after a PROGRAM, ERASE, or COPYBACK operation.
• Under typical-use conditions, a minimum of 12-bit ECC for MLC devices is required per 540 bytes of data.
• Use bad block management and a wear-leveling algorithm”.
 

AdamK47

Lifer
Oct 9, 1999
15,782
3,606
136
One thing I noticed is that the 512GB Vertex 4 in my array with the 8 bad sectors has the serial number starting with 2Y3U... The other five 512GB Vertex 4 drives have the serial number starting with OCZ-...

The drive appears to be from a very different batch.
 

AdamK47

Lifer
Oct 9, 1999
15,782
3,606
136
Just performed a chkdsk on the array. It seems OK after about a months use.

chkdsk-3tb-array.png
 

AdamK47

Lifer
Oct 9, 1999
15,782
3,606
136
A week later and no change:

ReallocatedSectorsAfter.png


I started SSD Sentinel every day since last week to get proper logging. Looks like the 8 sectors on this drive and the 1 sector on the drive below it has the bad sectors noted from manufacturing.

BTW - How do I translate the "Total Count of Write Sectors" into something understandable? The drive in the screenshot currently shows 3,006,165,257. Disk 2 in the same array shows 558,279,319. The 256GB Vertex 4 that I've had since September shows 259,787,698.

Not sure why each drive in the 3TB array has different write sector counts. The drives have been used in the same array and have never been used by themselves.
 

AdamK47

Lifer
Oct 9, 1999
15,782
3,606
136
Update:

Four months later and no change in reallocated sectors on any of the drives. Bought two more for 8 in RAID-0. One came with 1 reallocated sector and the other came with 2 reallocated sectors from the factory.

Nine OCZ drives and no failure? According to the brilliant people on this sub-forum, the odds of an OCZ drive failing on me should have occurred by now. ;)

4TB-SSD-HDSentinel.png
 

lavaheadache

Diamond Member
Jan 28, 2005
6,893
14
81
Adam, I have bought 6 or more OCZ ssd's over the last couple years and all are still happily ticking away. I think a lot of problems are user specific more so than product specific. Some people always have issues while others little to none.
 

xeledon20005

Senior member
Feb 5, 2013
300
0
86
Stumbled upon this thread ive had my samsung 840pros for about 5months now im going to install this utility and check and see if ive got bad sectors, running them in a raid 0 config 128gb each.
 

GlacierFreeze

Golden Member
May 23, 2005
1,125
1
0
It's a very nice app. Hardware.Info is actually using it in their live stream Samsung 840 endurance video.

http://us.hardware.info/reviews/4178/hardwareinfo-tests-lifespan-of-samsung-ssd-840-250gb-tlc-ssd

I don't think the thing has ran for a full day because TB/s written per day is at 7.25 and rising (edit: reached 7.65 and rising very slowly now, prob level out not much higher than that). Not sure what it will but at that amount per day, the chart shows Wear Count to reach zero in 9 months. Will be sooner when TB/s per day level out though. Still pretty impressive even for the TLC memory. Basically only thing to worry about is how reliable the firmware is for any specific drive these days...
 
Last edited:

AdamK47

Lifer
Oct 9, 1999
15,782
3,606
136
Yep, looks fine to me.

Just for the record, all 9 of my Vertex 4s are still going strong. Sorry to be the bearer or bad news to those who don't like OCZ. :)