Trials and tribulations with a Crucial M4 in a T420

[at both ends]

A report just in case it helps somebody else ...

A little over a year ago I bought a new T420 with a 160GB Intel X25-M because Lenovo was offering a great price on a discontinued SSD product line. The machine is dual boot between Windows 7 and Ubuntu 12.04, but usually with Linux running. It's been trouble free no matter which OS is up. Disk space was getting a little tight, partly because of the dual boot, partly because stuff accumulates. Best Buy offered a great deal on a 512GB Crucial M4 and I couldn't help myself.

Swap in the new M4, restore Windows from the recovery disks, all looks good. Shrink the Windows volume, create some new partitions for Ubuntu to live in, install 12.04 again. Copy everything off the X25 using an external enclosure. It's all good.

Until the machine goes to sleep, either under Windows or Linux, either from software or via a lid close. It won't awake from sleep. No keypress, no mouse movement, no touch pad or trackpoint movement, no lid movement will revive Sleeping Beauty. Worse yet, powering the machine off and then back on results in a failure to detect the hard drive (error 2100), not just by the OS but by the BIOS itself. The only way to recover is to power completely down - battery out, unplugged from the wall - and then remove and reseat the drive.

Google soon turns up many mentions of Crucial SSD code 2100 detection errors on attempted wake in Thinkpads. (Also, for that matter, OCZ SSDs.) If only I'd known to look for that error code before purchase!

So what's the secret to getting this to work? Update your BIOS and update the SSD's firmware. On a Crucial M4, you have to go at least to version 010G. The T420 has now been running for 105 hours, sleeping and waking without trouble. I think it's fixed. Think, because I'm still worried about it and keep turning up other mysterious reports of failures like this.

[mikemex]

I read somewhere that SSDs are going south pretty fast lately. Not only Crucial but intel too; there is a bug in the 320 series that causes the drive to lose capacity and become an 8MB drive. Newer manufacturing processes are reducing the life of the cells significantly; form earlier 100K-1M writes to just a couple thoushands now. They struggle with the wear leveling algorithms and it shows.

Best thing to do for now in my opinion is to return it and wait. Since you use it to store your valuable data it's not something you want to take chances with. Myself, I would keep it, but only as a secondary drive in my desktop computer where I can backup data frequently to my NAS (which uses regular platter drives). And only for non critical stuff such as music, movies, games, etc. Never as my main drive.

I'm using Samsung first generation SLC drives for that reason, even if it limits me to 64GB. I don't trust the newer MLC drives.

[ThinkRob]

I read somewhere that SSDs are going south pretty fast lately. Not only Crucial but intel too; there is a bug in the 320 series that causes the drive to lose capacity and become an 8MB drive.

1) The 8MB bug was years ago.

2) It's been fixed for almost as long.

3) It hasn't affected the last two generations.

In other words, no, SSDs are not "going south pretty fast lately". It's pretty easy to pick a drive that fits your requirements (compatibility, write endurance, MTBF, etc.) if you just do a bit of research. Same as any other component, really...

[mikemex]

Ahm... I don't like to discuss this sort of things because it's mostly a matter of faith more than reason, but let's just say that unlike a regular platter drive, where the algorithm to store and retrieve data is pretty straightforward, an SSD uses very complex routing to compensate for the unreliable nature of the flash memory and is more dependent on the quality of the software inside. With all those models of SSD coming out I am not surprised when they rush it to get a product out on the market well before it is properly tested. Besides, since it's solid storage and very dense, it's susceptible to the same quality issues (commonly known as "yields") as any other chip. Solid state parts physically wear: they suffer from electron migration and other issues. So even if the drive works fine out of the box, it might suddenly start to fail in a way the firmware isn't programmed to compensate.

Read this:

http://www.tomshardware.com/reviews/ssd ... ,2923.html

All in all, I'm not against SSDs, I'm using them in all my machines now. I'm just calling to be careful. I know MLC drives wear much faster from first hand: with about the same level of usage, an MLC drive had 50% life remaining (reported) compared to 95% of an SLC drive. Both manufactured by Samsung using identical chips. And that's with much older chips which were far more reliable than modern ones.

Don't fool yourself, SSDs are problematic, to say the least. Just do a quick search on this forum and you'll find all the issues related to SSDs, starting by compatibility. I've never read such things related to platter drives, say, someone starting a thread saying that drive X didn't want to start on a particular machine.

[ThinkRob]

I'm well aware of how SSDs work.

I'm also well aware of how hard drives work, which is why I'm not too worried about SSDs.

Both types of drives die. They will die, and they will die a lot. Back up, avoid drives at too-good-to-be-true prices (unless your data isn't important), and expect to replace your drives after a couple years. Same strategy I use for both spinning rust and shocked sand.

[jayton4]

Really the only tech website out there that reports on SSDs that has any clue of what they are talking about is http://www.anandtech.com/ Tom's Hardware is recognized as being one of the best tech websites (and the longest lasting that I know of) for product information. Covering SSDs is not their strongest area by far.

I have worked with and followed closely the SSD market since 2007. There is nothing wrong with the industry and they are not in any way going south. Wear-leveling algorithms have continuously improved with each generation of controller released. I have worked with, used, tested, and sold everything from rugged SLC based drives custom designed for defense contractors to just about every MLC based drive available in the consumer market.

The MLC based SSD will last for many years with average use. There are several things the user can do to accelerate an early death that are well documented on the net (disable defrag, disable page file, no torrents on the SSD, and so on). Manufacturer's will use NAND of different type and quality in different lines of drives to compete at different price levels which is clearly documented. The problems today with SSDs are related to the firmware of the controllers, not the NAND. These problems get ironed out over time with firmware updates. I wouldn't pay too much attention to google searches on which drives get what errors, because all of them have had errors at some point in their product cycles.

I almost lost a very large account earlier this year because of the 5000 hour firmware bug in the Crucial m4. The firmware was fixed sometime in March. My experience has been that the 0309 firmware is the most stable. Since then I have sold several hundreds of these drives to very satisfied customers.

OCZ was the first to market with new SF-2281 drives back in the summer of 2011. At that time OCZ was near the top of the industry in volume shipments with their previous generation SandForce drives. The new 2281 was riddled with problems that did not get ironed out until October of 2011. This badly hurt OCZ's reputation and I'm sure a web search for OCZ's problems will turn up a vast number of hits. Things are different now, and new SF-2281 based drives are very reliable and new models are released very frequently and several manufacturers have switched over to using SandForce SF-2281 exclusively. Corsair, Patriot, Kingston, Sandisk, PNY, Mercury, Mushkin, ADATA, OCZ, Intel, and many more manufacturers produce very solid and reliable SF-2281 based drives. I personally own 6 of the SF-2281 based drives myself and they are my choice, including for my two main ThinkPads.

[ThinkRob]

My bias against OCZ comes not from their products -- I don't have enough data to know whether there are systemic problems with those -- but from my and others' experiences with their warranty support. In my case, the process of getting an Agility replaced was such a colossal PITA that I never want to have anything to do with them ever again.

[mikemex]

I wouldn't pay too much attention to google searches on which drives get what errors, because all of them have had errors at some point in their product cycles.

Where is common sense this days guys? As an end user you souldn't be expected to deal with firmware issues and such. Either a product comes usable from factory or it is bad quality.

[bill bolton]

Read this:
http://www.tomshardware.com/reviews/ssd ... ,2923.html

Don't bother, its nearly a year and a half out of date

I've never read such things related to platter drives, say, someone starting a thread saying that drive X didn't want to start on a particular machine.

There are plenty of those threads on here from earlier times

[ajkula66]

OCZ was the first to market with new SF-2281 drives back in the summer of 2011. At that time OCZ was near the top of the industry in volume shipments with their previous generation SandForce drives. The new 2281 was riddled with problems that did not get ironed out until October of 2011. This badly hurt OCZ's reputation and I'm sure a web search for OCZ's problems will turn up a vast number of hits. Things are different now, and new SF-2281 based drives are very reliable and new models are released very frequently and several manufacturers have switched over to using SandForce SF-2281 exclusively. Corsair, Patriot, Kingston, Sandisk, PNY, Mercury, Mushkin, ADATA, OCZ, Intel, and many more manufacturers produce very solid and reliable SF-2281 based drives. I personally own 6 of the SF-2281 based drives myself and they are my choice, including for my two main ThinkPads.

Well...OCZ and Mushkin kept on shipping drives with debatable firmware which kept on failing *long* after October of 2011. As a matter of fact, they are STILL failing. Just read the reviews on NewEgg from the folks who bought them recently. It's a disaster.

I couldn't care less about OCZ after an arrogant and ignorant response I've gotten from them after complaining about the size of one of their Vertex drives and hope that they collectively kick the bucket as a manufacturer. Feel bad for Mushkin though, whose RAM offerings I always liked and still use in several machines.

All of that said, I wouldn't install anything with a Sandforce controller in any of my machines. There are many other - far safer IMO - choices around.