Possible reason to avoid OCZ?

I know that OWC is a competitor to OCZ, but this info is pretty damning if it's true:

Not All SSD?s Are Created Equal: The Story Continues | Other World Computing Blog

"When we took the cover off of this third, direct from OCZ SSD, we found a ‘S’ stamped over Micron logo on all the flash devices (see the image to the left). This indicates the device is “off spec” product because it failed some parameter of Micron’s full performance and/or quality specification testing. “Off spec” memory is typically used in low-level applications such as toys, offering considerable cost savings over Tier 1 level to an SSD manufacturer.

As OWC only uses (and would only ever consider using) Tier 1/Grade A chips in our Mercury SSD models, an inquiry was made with a Micron product representative on their thoughts on the use of off-spec flash memory in a Solid State Drive application.

“It is a very brave action to take, using these chips in a data storage device,” was the reply given."


*** Update *** OCZ's reaction

Guide 34nm Spectek Nand...see here for specs and info

 

From what I know of the company, this doesn't surprise me at all. Because of all the legal fun that will probably ensue, I certainly hope that OWC can back up their claims. Otherwise the legal judgment against them is going to really, really hurt.

 

Wow, I sense a disturbance in the Force..........

 

Good, with all of this bad press maybe they'll lower prices and I can get a SSD =p

 

not surprising me much either...

fortunately its easy to prove this... buy a SSD, open it. and show to everyone...

 

What they need now is a bunch of third party reviewer sites revealing the same information and then have the flash manufacturer come forward to actually confirm the 'S' thing in their own article.

 

This is why DuraWrite, DuraClass and LTWT is present in SandForce products: the easy use of 'garbage' nand, greatly exaggerated 'up to' speeds and real world performance that is below mechanical HDD's (thanks to LTWT, of course, to protect that subpar nand used).

 

I mean I thought there are very few actual manufacturers of flash NAND, so wouldn't the controller matter more so than the binning of flash NAND?

 

Well, the part that makes we worry is does it mean that the NAND is more likely to fail than Tier1 quality? I'd like to know what tests it failed.

 

This looks like a lot like a bad smear campaign to me... Never heard much of OWC, so their "claims" of being the best seem kinda fishy. Haven't found many reviews on them either that proclaim their immense reliability or performance. Not that I'm an OCZ fanboy, the whole 25nm fiasco was ridiculous, but I think a lot of this could be cleared up if people with vertex 2s opened up their ssds and looked.

 

The fact that NAND fails any test is bad. It's like using RAM that doesn't pass MemTest86+ if you ask me.

Go check out specification sheets for NAND and you'll find a lot of electrical engineering chat about timing requirements and stuff like that, not to mention estimates on data durability v. environment and all that. This NAND could violate any one of those specs.

My question at this point is if this is the NAND they used when they did that exchange program from the 25nm debacle. They go for the lowest cost solutions in almost every product (hence secretly going with chips that use lower program/erase cycle ratings and then secretly going with 25nm without changing the product's model number).

It is about time that someone demand the government investigate OCZ for potentially illegal business practices.

 

OWC is a vendor NOT manufacturer last I checked (if OWC wishes to come prove me wrong, I'd like to know where they are manufacturing their circuit boards, and housing as we known everything else is OEM OR at the very least where they assemble all their parts.) They are mostly known in the Mac arena due to their e-sales website.
They run by their more commonly known name: Macsales.com

I would not be surprised one bit if they actually buy their SSD already assembled/packaged from one of the bigger SSD companies.

 

hmmm... well I know my Agility is the last OCZ I buy... I don't like companies that are decpetive, it's to bad I've had lots of really good luck with them. I'll be looking a G.skill... they were good about my last SSD when it died.

 

There is nothing illegal other than the stated size is different than the actual which may be very close to border line case.

And no, OCZ is too small and no political value to chase after.

 

They say in teh comments:

"we build with Micron, Toshiba, intel, Samsung, and Hynix depending on model and such.

A good majority is Micron."

 

They dont claim to be "the best" it says right on their SSD site: One of the fastest Internal SATA 2.5" SSDs available.

Which is true.

 

No no thats not what I meant.
Let me explain how most manufacturers work.

OCZ buys NAND/Controllers/Etc. They manufacture or OEM the circuit boards, controllers for sata etc and housing. Then they assemble it themselves usually via a factory, package and ship.

My concern is that OWC is more then likely paying a larger company to do everything OCZ doesn like getting all nand, controllers, housing, assembling and then packaging for OWC. This is called reselling, or rebranding and is usually done by vendors.

 

The only hint about where they are assembled is that they say, "Built in the USA from domestic and imported parts."

And: OWC assembles and sells its own line of hardware products under the Mercury™ brand. The products include accelerators, hard drives, CD/CD-R/CD-RW and DVD/DVD-R/DVD-RW drives, memory, FireWire accessories, storage device enclosures and leather accessories.

 

Aha so maybe this is why OCZ drives start performing bad (lower speed than specs) after a while? These flash devices that have failed Micron`s tests may fully function at first but start withering after a while because it is something wrong with them/bad flash?

If so this is huge!

 

Using NAND that flash manufacturers state is not fit for data storage might fall under false advertisement as well (since it is being marketed as a storage device).

No, that's more consequence of using flash. After the cells are written to the first time, all subsequent writes have to erase the cell first too. This affects all drives regardless of brand.

 

except it doesn't in the same way, greg. on samsung and intel, you don't feel the impact much.

other than that, ocz is dead for me anyways since years (i won't forget their core debacle, and all the newer crap they do just fits the picture i got from them back in those days, this news included)

 

So why do Intel drives perform so much better and is able to sustain the speeds that they are spec`ed out to be?
It has nothing to do with the 0 and 1s since reviewers see brand new OCZ SSDs fall way under the specs after a little while.
Very fishy to me

 

Huh ? There is not such law saying whether something is fit for data storage.

 

Correct however if massive failure rates occur, one can then dictate that it was not suitable for the purpose. This would be a comparison of lets say Faulty SSD A, vs the rest of the industries offerings.

 

There isn't a massive failure rate though. It is a rate that is not acceptable to me but not to the point that would cause it legal trouble(i.e. not breaking any law).

Beside, it has so little money in the bank(their financial situation is simply terrible) and not the kind of company where those legal sharks would smell blood for class action law suit.

 

True true.

 

A French hardware site ran some statistics of failure rates of computer components a few months ago. OCZ's SSDs (which had the highest SSD failure rate at just under 3%) was still significantly more reliable than any mechanical HDD (most had failure rates of about 5%, the least reliable, which I think was a Caviar Black 2TB, was about 10%).

There isn't any real legal risk for OCZ even if the lower-grade NAND has a significantly higher failure rate. After all, they can afford to more than triple their current failure rate before they start beating out HDDs.

 

French Retailer Data Offers SSD Failure Rates | News | The Mac Observer

 

question is, will the rates be the same two years from now? how long do they last? i know my intels will last for ever, more or less (the internal lifetime rating minimum is 5x higher than the one written on the box, and that lifetime is long enough to never matter).

hdds have higher failure rates as their failures happen randomly over their ordinary lifetime. they life for ever or till they randomly die.

an ssd doesn't die randomly (much), but they will die after some specific amount of time, more or less all of them. now using lowest quality flash will make them die much sooner, obviously. so the question is, how long will they survive?

 

i bougth an ocz 120gb 2e earlier today and cancelled after doing some additional research. THANK YOU NOTEBOOK REVIEW is all i can say! I've put an order in for the F120 corsair SSD, picked it it up for just under £170. Anyone had any issues with this one? I've made sure (as best as i can) that its not the F115. The one ive picked up is s equivalent to OCZ's old (and better) Vertex 2e, right?

 

Every drive experiences this. It's something you have to put up with to use flash. How well the controller hides this from you (by proactively erasing cells if it is aware of the file system or used blocks, etc) is the difference that you ask of.

You have to clear the cell to use it again and to do that it takes time.

See comments below.

It could also be argued that since the flash manufacturer (who?) said that the flash wasn't fit for data storage, and OCZ claims it is...

At least not yet. I've read a decent amount about users having their drives die every year or so and while I fully acknowledge that that may or may not be any indicator of OCZ's potentially non-existent problems...I have to wonder if there is a grain of truth there. I myself had nothing but problems with their drives all the way back to the Agility '1' lineup.

Their mismanagement and their careless choices with SSD manufacturing (read: flash quality, PCB quality, firmware quality, etc, etc) are probably going to be what kills OCZ as opposed to a class action. That being said I'd love to see the legal fireworks that might come out of OWC's article.

Sad to say that I've got 12GB of OCZ RAM floating around here, and the moment it starts having problems I know that I won't be able to get the promised warranty support (they existed the RAM business for all intents and purposes and there are plenty of complaints on their own forums about RAM RMAs).

 

greg: if you have trim, you don't need to experience it

 

Let`s say you are running a program with the SSD. It is a new SSD and is filled around 5% of it`s capacity. Doesn`t the drive fill unused blocks instead of deleting and rewriting on a used block? The reviewers are using brand new SSDs which should have no capacity related speed issues.

Are you saying that the controller of Intel drives are much better at erasing cells than the SF controllers and the Intel SSDs will therefor be much better at holding the speeds? Sounds valid

 

No one knows the true reason why SF drive would get slower down significantly once enough data has been written. According to OCZ(if you believe them), it is because SF deliberately slow it down to protect the NAND. It is also possible that it is some design limitation. SF drive are compressed. This is similar to NTFS compression which is well known to result in much more segmentation.

 

Aha ok i see. Comparing Intel and Sandforce drives, why do SF SSDs need to slow down to protect the flash memory while Intel don`t have to?
If it is because of that SF drives have faster speed in general compared, what is the point if they have to slow down anyway?

 

No idea, that is the line OCZ touted. I am a bit skeptical and incline more to the compression issue. Only SF drive seems to not responding well to TRIM.

Though the rumour said they 'lifted' the speed break in SF2XXX line. Again, no confirmation of real life usage data.

The SF1xxx line IMO is a hype. When it is new, sure it is faster but I am not too sure if those numbers shown by the 'reviewers' are done on dirty drive(AFTER TRIM), which is the true usage experience most users would experience.

 

The TRIM problem is entirely Sandforce's fault - they implement TRIM support in such a conservative way that there's essentially no difference between having TRIM enabled and disabled.

Fortunately for Sandforce and their partners, nobody pays much attention to the slow dirty drive performance because it's still faster than Intel's. Then again, G2 sequential write speeds are so slow that they're not hard to beat.

 

One thing I am not sure is if it is they being 'conservative' or 'unable' to handle it in an efficient way. They said that is a choice, I have my doubt. The reason is that handling segmented file system is not an easy task. There are always trade off to or not to do it.

Most people don't pay attention to because the slow dirty drive is still very fast(comparing with say HDD) but whether it is faster than other SSD drive, I have no idea and very few people have really prolonged real life comparison.

 

They have to slow them down because of the subpar nand they use inside.

 

Yeah that is what i have been asking all along. Is their NAND any worse than the flash memory other manufactorers are using?
Or is the non sustainable speed of OCZ SSDs due to bad controller?
Or because of compression like chimpanzee told me about?
Or all above in a terrible mix?

 

i know my flash chips have no marks on them, in none of the ssd i've opened.

 

What brand are you using dave? A little alarming that the OCZ drives had a big S on them saying it didn`t pass the tests that should ensure reliable and working SSDs

 

OWC which challenge OCZ's NAND used(and said they use only high grade NAND) also use SF controller and suffers from the same DuraWrite speed break.

 

If you go to AV and HTPC enthusiast sites, you'll find tons of real-life uncompressible write comparisons between SSDs. Dirty Sandforce drives are still able to write h.264 videos at about 110 MB/s, so long as they're 100 GB or larger. This is marginally faster than the X-25M's 100 MB/s.

That said, while this minor victory is enough to deflect attention away from the problem, Crucial's C300 is the clear winner in this area with transfer rates well in excess of 200 MB/s.

 

intel g1, intel g2, samsung whatever old gen (they're 2, 3 years old now? one has 200MB/s read speed, one has 100MB/s). a c300. some mtrons.

 

Intel chips I would expect to have a big i on them, they don't use the intel CPU logo on their NAND, just a big "i"

 

at that rate, I am not sure I would pay for the SSD price. A fast spinning HDD can exceed that speed and they are much cheaper in terms of $/GB

Personally, I would just use fast HDD(like velociptor? or those 3.5" equivalent) if I want fast sequential read/write. Using SSD like that sounds to be a waste to me.

 

Honestly, I thought the same thing until I got a cheap Kingston 128gb SSD on woot one day. Then I installed it and I could never go back, from now on everything I own is going to have an SSD or get an SSD upgrade.

 

Do you use it exclusively as a 'scratch disk' for say photo/video editing purpose where sequential read/write is important or you use it as a general System/OS disk ?

 

it can only exceed that speed in one rare case: where everything is perfectly sequencial. the moment one thing accesses any other part of the disk (like, about anytime), the ssd will still win.

do you have an ssd yet?