Arguably, the two most important components of an SSD are the controller and NAND flash memory. In the memory industry, manufacturers may dispose of sub-par memory on the open market where, at least in the case of SSDs, more suitable applications may be found.
The potential downside to this commonplace practice is the memory has the potential to be counterfeited and sold, while often suffering performance or capacity concerns.
Up until this report, we have yet to see any instances of counterfeit SSDs. This analysis provides pictures, identification and full analysis of a fake SSD, offering the consumer a tool to assist should they ever believe that their SSD ‘just isn’t quite what it seems’.
INTRODUCTION
As much as counterfeiting hasn’t been seen in the SSD industry (at least not that we could find), it is commonplace with flash drives. The Internet is littered with instances of EBay purchases where an incredible deal was indeed ‘too good to be true’. The common thread between the two is that both contain NAND flash memory, this component being the main culprit with flash drive forgery.
As much as it seems a bit unusual that we cannot locate prior instances of fake or counterfeit SSDs, we would never have predicted our own receipt of a counterfeit SSD in a sealed package, and from the manufacturer directly. We guarantee the manufacturer never expected this review.
Can you identify the fake? Click on any of our pictures for a higher resolution shot.
DISCLAIMER
This report involves the SSD manufacturer KingFast, as well as Micron and OCZ with respect to NAND flash memory. It is our opinion that all three companies are victims of this occurrence and had no knowledge of, or intent to produce or distribute the counterfeit SSD received. All three companies have been contacted for background and assistance and have been very accommodating in assisting with this report.
BACKGROUND AND INITIAL CLUES
In January 2013, The SSD review received a KingFast F3 Plus 240GB from a KingFast representative. On receipt, the exterior package and the interior clear plastic SSD were both sealed. Within the exterior packaging of the KingFast F3 Plus SSD was a certification sticker, acknowledging that it had been inspected prior to leaving the factory. There was an additional seal on the side of the drive that also acknowledged inspection of the F3 SSD.
Traditionally, we might take pictures of the SSD before testing but our first clue as this SSDs characteristics became apparent when we saw the performance test below left. In assisting us with this report, KingFast has sent another F3 Plus SSD and its result for the same test is below right. It was after this that we elected to do an Internet search on other similar KingFast reviews and we discovered that, regardless of SSD model, all KingFast SSDs utilized Intel memory exclusively.
For those new to the game, AS SSD tests with highly incompressible data samples and is the bread and butter for determining write transfer speeds when it comes to incompressible data, the data type we most often see in movies, music and photography. This metric is key to so many industry professionals migrating to SSDs because, simply put, faster data transfer equates to more work being completed in a smaller time span.
In today’s SSD arena, high sequential read and write performance of upper tier SSDs (which this should be) can be expected to be somewhere in the 300-500MB/s. When we see results of 213MB/s read and152MB/s write, alarms start to go off.
Things weren’t much different when we took a look at the AS SSD IOPS and then conducted an AS SSD Copy Bench Test where, once again, performance deviations between the same model drive were very evident.
Excellent job as always Les. Great investigating skills 🙂
Fake Crucial ssd’s turned up on the Crucial forums about 4 to 6 months ago, the older M225 ssd’s and at least one M4, I placed a caution in the news section of my thread at extreme overclocking, These ssd drives were not manufactured by Crucial and were discovered when the ssd’s were rma’d, the M225 is a old Crucial drive that’s been EOL for a while now, the rma’s were naturally turned down by Crucial and a warning was placed in the Crucial forums.
Would love to see a link for any piks of those SSDs PCB. Spent hours looking for exactly that when compiling the article without luck.
Could you share the link? All I can find are threads where people think the “new” M4 box design is fake..
Would be interesting to find out if it’s possible for an ID/serial# to be burned into a NAND chip during manufacture. The controller would then verify the chips ID at power-up.
Love the CSAM image! That really puts the ID on where they came from originally!
Wow! I’m stunned! Good read too! Gives important to benchmarking hardware we buy and see if it is up to par to what it is claiming.
That is Agility 4 Asynchronous NAND that didn’t make the cut or was resold by OCZ. What most likely happened was that Kingfast put an ASYNC board in an SYNC (or just the wrong sticker) enclosure by mistake. This was intended to be sold as one of their ASYNC drives which would have been legit except for the consumer not knowing that re-branded and potentially inferior NAND was utilized although it may not even be NAND that didn’t make the cut it could just be good Async NAND that OCZ sold as surplus to raise capitol.
Credible thought but the memory in use was fraudulently marked as Micron memory which would be synchronous, although never having been used in an SSD prior. Kingspec informed us that the belief was that they were purchasing suitable memory for the F3 Plus.
That memory is async and OCZ async memory is micron
Ok, I have to ask why you might feel this is Agility 4 Async? To start, when a wafer is purchased from Micron, there is no separation from sync/async and this becomes part of the build and packaging process. For any OCZ purchases where they package their own memory, their purchase from Micron (or any other) are not sync/async wafers. Next, the high sequential and 512 write performance of incompressible data is far worse than our testing of, not only the Agility 4, but any drive we have tested to date. Did you see matching numbers on an Agility 4 or can you provide a bit of background?
Les, the Agility 4 uses Micron Asynchronous 29F128G08CFAAA nand memory, the numbers slightly different, this numbers from a 256gb drive.
Yes we were aware of that, however, the numbering is counterfeit. It does not match the internal make up of the die, but to say that it is 32GB (ie 256). Remember, when one purchases a wafer, as they did here, that wafer hasn’t been determined to be sync or async at the point of sale so nobody could qualify such afterwords. We can attest to the fact that the counterfeit Micron LABEL does not match the die within.
I’m quite sure your right about the fake nand, the fact that it’s the OCZ numbering that’s been blacked out and not the Micron number would mean it as little do with the Agility 4 as they retained the Micron number even though the number does actually exist, not sure about wafers, but the the actual chips are ordered as either Asynch or Synch, there’s a third option as well, I only know that the triple A stands for Asynchronous memory and those 256gb chips may be used on the 512gb Agility 4, all I’m saying is Bill might be right at least as far as the numbers concerned.
With the information available, there is no way of knowing the specifics of the memory we are discussing, including whether it is sync or async. When wafers are purchased, there is no discussion whatsoever of whether wafers will be synchronous or asynchronous, they are only labeled as NGD wafers. The decision to produce such or asynchronous memory only occurs in the packaging process when the memory is built.
Thanks Les, I should add this was a great article and I am interested to know why some memory is sold as Asynchronous and some, the better Synchronous memory, nand memory differentiates a good or not so good ssd drive, it and the controller are the main things I look at when checking out a ssd drive, so more the I know about the memory within a drive, the better I can assess how good the drive is, or likely to be, looking forward to your article.
It’s pretty obvious by the tests that the memory is sub-standard and fake, even poor asynchronous memory performs far better than the benchmarks you released. Thanks for the info Les.
Wow! I really appreciate the investigative work you did for this article. Knowledge is power, and you taught us what to do if we have doubts about an SSD recently purchased. Thanks man!
great and thorough Article!
Outstanding article. I give you the utmost respect for giving each of the vendors a chance to provide information before posting this story. There are many sites that would have just blasted them as selling fakes since they are a lesser know company. Great Job on looking for the truth instead of just a story.
Crazy, Les. Great catch though! SSD Sherlock Holmes.
You caught big fish this time, that bastards always try to scam the consumers ! Great Job !
Story is popular. Eteknix repackaged it. https://www.eteknix.com/kingfast-unknowingly-sent-counterfeit-ssds-with-mislabelled-flash-nand-for-review/
Yes, we have no less than 20 Asian sites that have grabbed it as well (of course breaking copyright with several of our pictures), but are nevertheless glad to see that it was a cause well worth fighting.
This poses an interesting quandry when SSD pcbs are under sealed covers with warranty seals on them.
Did the snowblower (and screwdriver) survive?
Yup, but there was a point when a sharp scream could be heard a block away.
a Very-very Good Article (y)
Thanks for bringing this up (y)
very informative… everyone should read this.. for additional information
“…a concern may be evident in SMART IDs 01, C3, c9 and CC where read errors seem to have occurred.” I would not be concerned with these results in CrystalDiskInfo. I have 2 SSDs in my system with the same or also same figures (and no, they are not counterfit, I tested them)
Thanks for the input! Appreciated.
to be frank, this is sort of “norm” in this industry. someone “junk” can be others “treasure”.
Some guy over [H], who told me he used to work for PNY, said they’ve done this before. I told him that he was BS’ing me and that a company like PNY wouldn’t be doing such thing. Then he went on to tell me that it was my money and that I was wasting it. I wasn’t though as I have a pair of Deneva 2R SLC’s that retail $1450 each LOL.
That said, KingFast has a load of RANDOM drives on eBay. Just look up for KingFast and you’ll find many colorful and silly pages of SLC SSD’s being sold…
and those SLC’s are mostly based on that old Micron JMF controller with the shuttering bug. You know, the first SSD controller that got into the public back in late 2007.
They still sell proper and nice SLC drives though they’re all SATA2 drives based on old Indilix Barefoot controllers.
Checked there just now and not one. Can you provide a link?
Sorry, they were Kingspec, not Kingfast:
https://www.ebay.com/sch/i.html?_odkw=kingspec&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313&_nkw=slc+ssd&_sacat=0
You know, like those ROSS headphones they put next to KOSS headphones on Radioshack… lol
Yup thanks for clarifying and if you do find similar, let us know! We just may buy one eheh.
Huh, that micron NAND is definitely Async Only.
https://www.micron.com/~/media/Documents/Products/Part%20Numbering%20Guide/numngnand.pdf
The last A in the product number means it is Async Only.
This is kingfast playing loose and fast with it’s NAND control, not even checking what it is soldering onto the board, and the sandforce happily using whatever it is given.
I was worried the NAND was fake, and we would start seeing strange barely working solid state drives like we do for counterfeit USB chips
Also:
a) I should read reviews before commenting. I see how they are faked now … however the micron marking is correct. They have been marked Aysnc only.
b) For the SMART IDs 01, C3, C9 and CC on sandforce drives, this raw counter is some kind of internal error rate tracker. It will be always non-zero on sandforce drives after they have been reading/writing for a bit. Even drives with high quality NAND will show non-zero here. The important number is the normalised number of 01, if that falls below 50, it means the drive is having to correct too many errors internally and will likely show external errors.
So much technical stuff. So, do we buy Kingfast SSDs or avoid like the plague?
KingFast has not encountered any similar problems since this posting 2 years ago. They are still going strong without issue.
Kleftico-
This is an interesting story and string. If you had not commented on it I would probably not have seen it since it is over two year old.
I am beginning to come to the conclusion that one should probably try to get an “enterprise” SSD; anything else is a crap shoot. Or go for a cheaper SSD and just plan on upgrading again in a few years.
This is an excellent article and investigation that went as far as it possibly could; the players – KingFast, Micron, and OCZ – should have investigated this further at the time and identified exactly where the procurement breakdown occurred.
Jim
While it is true that enterprise SSDs will have a BOM, remember the fact that the memory here was blackballed to be seen as a higher grade; this could have hapened in consumer or enterprise.
Further, Kingfast did institute better QC immediately after this occurrence and it was limited to a single batch that was recalled. We have yet to see this duplicated.
Les-
I agree that this is primarily a QC issue but it appears to me that consumer level drives aim for lower performance and lower costs, which both imply lower reliability and quality control. Just look at the number of last years’ heroes that are this year’s goats. If KingFast had not sent a bad one to you, I wonder if this would have been known.
I don’t think Cadillac, Lincoln, or the NSA (National Security Agency) put up with the levels of performance and reliability that are acceptable, it seems, for consumer SSDs even now.
Jim
Yes I would agree that the BOM requirement of enterprise drives ensures a much better QC because the components arent open to be changed without recertification.
Les – thanks for your reply. Ordered one. And thanks for the article.
Jim, – Glad I did then. Fascinating – a right royal cockup on their part. Red faces at Kingfast. I came here while seeking reviews for Kingfast SSDs as I had my eye on one. Les replied to my query, “KingFast has not encountered any similar problems since this posting 2 years ago. They are still going strong without issue.” That and my belief that they are still going to be vigilant – even though it’s two years ago, whoever placed the order will still be getting ribbed about it – allowed me to cast caution, and I’ve just ordered one on Amazon – 240Gb. The reviews were all very good when price accounted for, but even without the price factor, the reviews were still all good.
Coincidentally, Giveawayoftheday, have today offered full program Minitool Partition Wizard Professional for free, and it contains a menuitem for transferring current OS to SSD. How’s about that! Downloaded, installed, and awaiting delivery of the drive.
I’ll get back when I’ve used it for a while.
Great news and thanks for relying on us for your decision. It is nice to get the support and know we are helping.
Even if this discussion is quite hold, I would like to add a new note: some time ago I bought (from cartft.de) two Kingfast SSD “wide”, that means they are guarantee to work from -45 to 80 degrees. That feature doesn’t appear on any sticker on the disk and, opening it, I’ll find a Toshiba NAND chipset TC58TEG6DCJTA00 that, from their datasheet, is the 0 to 70 degree model, the “right one” being TC58TEG6DCJTAI0. Now, the seller, CarTFT, says that that model, 2701MVS, is a wide temperature model but the fact is that it’s not. Who’s frauding? CarTFT or Kingfast?