OCZ’s 25nm Vertex 2 Performance Issues

Logo Courtesy: OCZ Technology

Logo Courtesy: OCZ Technology

There has been a recent uproar in the tech community regarding OCZ’s new 25nm Vertex 2 SSD drives. Usually we’re quite excited to see newer fabrication technology reach the shelves, as this usually results in a superior product. Sadly, this wasn’t the case with OCZ’s latest iteration of their Vertex 2 SSD offering.

Looking at some straight-forward benchmarks posted by StorageReview.com member TSullivan, the performance is what you would expect from a Vertex 2 SSD in your system – the image below left shows a benchmark of the 34 nm OCZ Vertex 2 60 GB SSD. The image on the right is a screen shot of the CrystalDiskMark benchmark for the 25 nm Vertex 2 60 GB SSD.

Image Courtesy: TSullivan

Image Courtesy: TSullivan

Image Courtesy: TSullivan

Image Courtesy: TSullivan

Whoa! What happened here? The write speeds are half as fast for the 25 nm version, while the read speeds are up to 30Mb/s slower (4K QD32) as well. That’s a huge decrease in performance. What’s worse is that OCZ and their online retailers (Newegg, for example) are listing the “new” Vertex 2 drive as having the exact same specifications as the “old” drive, when that’s clearly not the case. These two drives look exactly the same on the outside, until you open them up to see the visual differences, as shown below from StorageReview.com’s full review.

Image Courtesy: StorageReview.com

Image Courtesy: StorageReview.com

This is a developing story, so hopefully we’ll get some positive news from OCZ and retailers about how they are going to deal with this issue. Needless to say, the tech community is very unhappy about being sold an inferior product that is packaged identically and claimed to be the same as it’s superior counterpart. There are a number of  online communities discussing this issue.

Overclockers.com discussion thread

StorageReview.com discussion thread /StorageReview.com’s Vertex 2 Review

MacRumors.com discussion thread /OCZ’s Forum discussion thread

Update 2/16/2011 3PM: Official statement from OCZ — “[C]ustomers are welcome to go through our service program to exchange for a 32gbit model if they have concerns about their 64gbit drives. The new part numbers will reflect the 32Gbit die solutions to distinguish between the models (last page): OCZSSD2-2VTX60G, OCZSSD2-2VTX120G (non ‘E’ part numbers)”

OCZ Technology Group Official Statement – Driving down the cost of SSDs As the industry transitions to the 2Xnm process, OCZ is notably the first to market with this technology that aggressively reduces the costs of our consumer-grade SSDs. Most of our customers are aware that as the NAND flash technology process nodes are shrunk, the price of NAND comes down substantially. OCZ continues to focus on delivering the highest performance and highest reliability drives available at a lower price point…paving the way for SSDs to become more accessible to the complete range of consumers and to take the place of traditional mechanical hard drives over the next few years.

Due to this natural transition to next generation NAND flash components, certain 2Xnm-based SSD configurations may see a slight difference in the IDEMA (International Disk Drive and Equipment Materials Association) capacity, particularly lower density drives, due to the higher 64 Gbit die density of 2Xnm flash solutions. This is due to the need to reserve additional space for the drives’ sophisticated performance and reliability features which provide real world benefits and are not offered by many other manufacturers. These features include RAID-like data protection and recovery in the event of flash block malfunction, as well as advanced wear-leveling to enhance the SSD’s endurance and lifespan.  Due to the use of higher density chips, the quantity of blocks reserved for this functionality doubled as we reserve a single plane of flash for additional redundancy.

Many users familiar with SSD technology understand that over-provisioning flash is designed to increase, not only the life of the SSD, but the performance of the SSDs; however, OCZ is sensitive to issues particularly related to RAID arrays where capacity alignment becomes critical for optimal functionality.  Even the smallest variance in size can create issues. In order to help resolve issues such as this and to clarify the situation we are disclosing (for vendor ordering purposes) the part number significance for our SSD capacity configurations. While we cannot guarantee (primarily due to price variance) that all of these parts are available at retailers, we as a company will extend our efforts to make these parts available throughout our Global channel. In addition, we’d like to point out to our customers that these products have always remained in production and are readily available directly from OCZ.

We should note that this only effects smaller capacity drives with the “E” series of firmware. Non “E” drives have no capacity variance and higher capacity drives also have no variance.

For those customers that have already purchased an “E” part and wish to move to a 32 Gbit die based drive solution, OCZ will offer a program in which customers can trade in their “E” drive and receive a credit towards the more expensive 32 Gbit die-based drives. Customers only needto pay the difference in the raw NAND price and OCZ will ship the replacement drive once the original “E” drive is received. If you are interested in an exchange please contact an OCZ
customer service representative and we will be happy to assist you.

OCZSSD2-2VTXE60G 55 64
OCZSSD2-2VTXE120G 118 128
OCZSSD2-2VTX60G 60 64
OCZSSD2-2VTX120G 120 128

All 2Xnm-based SSDs carry the same warranty as previous OCZ SSD products and it is OCZ’s objective to give customers the very best balance of performance, reliability, and capacity ensuring an optimal experience for consumers across the complete spectrum of applications.

Update 2/16/2011 9PM: The latest information we received from OCZ states they will replace any of these particular drives for all customers who may be displeased with the product. Most importantly, they will cover shipping costs both ways and they will not charge for the pricing difference. In full, the following is the latest information we received.

Thanks Matt, appreciate you getting the word out to your readers :) We really want to make it right for people that are displeased, so if anyone is not satisfied with their drive they can contact us to replace. We actually will be covering shipping costs both ways and not charging for the pricing differences, that has just been updated from our support staff.

Regarding the performance, these drives are able to meet our published specifications with the same benchmarks we have always used, which is ATTO for maximum rated sequential speeds and IOMeter for random IOPS. We have also recently added a dual spec to show performance in ASSD on our product pages, we realize every benchmark is different and handles data differently.

Either way, if anyone feels their drive is not what they paid for we will replace it with as little hassle as possible :)



Just a nickname's Avatar
I heard the Agility 2 were also touched by the issue.

I wish OCZ will fix this issue soon Not cool for all customers who bought these SSD not knowing they would never get the listed performance!
Marshmallow64's Avatar
Great article. And yes the agility 2 were also affected.
I.M.O.G.'s Avatar
We've contacted OCZ for comment. Has anyone seen them make any official statements as far as the company's awareness or stance on the issue?
Krusher33's Avatar
Excuse the noobness of this question but why is one showing 56GB and other 51GB?
David's Avatar
I'm currently in contact with Newegg over any necessary changes to their advertising to reflect the issues with performance - some people will be sorely disappointed.
ratbuddy's Avatar
There's a sticky on their forums by moderators who I believe are employees. They downplay the issue with a mix of arrogance and dismissiveness.
Xaotic's Avatar
It relates to their RAISE implementation and overprovisioning to allow for wear leveling. There are also two versions one using 64Gb and one using 32Gb NAND ICs in 25nm drives. The old version, 34nm, uses only 32Gb NAND chips. The new versions could show either size based on the chips used. The 64Gb chips overprovision based on 8GB of OP space and the 32Gb OP based on 4GB of space.

OCZ will be releasing a tool to determine density and which die size is used on the chips, if what I have understood from the thread in their forums is correct. Though it could also be read as simply which die size is used.
Neuromancer's Avatar
Excellent coverage. Was reading up on this when I found it here.

Getting OCZ rep to comment would be great, I am sure they are a little frazzled right now though considering the backlash they went through on their own forums though

Yeah this is a result of moving to 8GB chips (missing partition size) and half the number of chips (RAISE reduction)
splat's Avatar
updated with official statement from OCZ
Xaotic's Avatar
Sadly, their comment fails to address the performance related question...
MattNo5ss's Avatar
To me, it sounds like they're saying the over-provisioning is causing the performance to decrease, but the over-provisioning is worth it...
I.M.O.G.'s Avatar
If it addressed the performance question, it certainly wasn't very clear. I feel that question is what is generating most of the attention and if the source of the performance issue can be explained directly then a lot of the community ruckus would die down.

Clarification has been requested.
ratbuddy's Avatar
The performance is down because they're using half as many flash chips with twice the capacity. Less chips equals less channels to write to at once equals slower performance.

The overprovisioning thing just relates to the smaller capacity of the drives compared to the older. The scheme they use calls for a whole plane of the flash to be dedicated to overprovisioning. In this case, I believe each chip has 8 planes, so you should see a loss of 8 gigs (64/8) on a 64 gig drive compared to a loss of 4 gigs (64/16) on an older model. This is just speculation/guessing really, because I don't know which flash they're actually using. I'm basing my numbers on stuff I remember from a few years back, so if someone know more current info about flash on the market today, feel free to correct me.

They really should have released the gimped drives as another model name, and now that they're caught, they darn well better charge less for the "E" drives.
Krusher33's Avatar
Oh bother... I'm one of those that purchased one of the drives listed by OCZ that is affected.

Tiger at the OCZ forumns state that a tool will be available to see which NAND your drive has. I think I'll wait on it before asking for a replacement.

Source: http://www.ocztechnologyforum.com/fo...OCZ-SSD-drives
Just a nickname's Avatar
I read somewhere that the firmware installed on your drive out of the box could indicate you if you got the damned version.
Dooms101's Avatar
Wow... I just checked my box from my Vertex 2 60GB I bought before Christmas:


I am going to see if I can figure out which one this is since the "E" can actually be either 34nm or 24nm versions. The best way to tell is by the format space. 34nm verions are 56GB and 24nm are 51GB.

Edit: well good news is that the 24nm versions shipped with at earliest 1.27 fw verions (most come with 1.28, source: OCZ admin). Mine currently has 1.25 so I am running a 34nm version.

Also, OCZ announced that upgrading to the x16 24nm versions are now free.
I.M.O.G.'s Avatar
Latest update made to the article - OCZ will cover shipping costs both ways for anyone unhappy with one of these particular drives in question, and they will not charge anything for pricing differences. Read the end of the article for the latest word direct from OCZ. They also directly address the performance question.

I still don't care much for sticking with the same product name in light of the nature of these differences, but that is common practice in this industry unfortunately. Most importantly to me, OCZ is doing what it takes to make this right for anyone with the product, who is aware of the issue... That means our audience is taken care of, and I'm happy with that.

Check your numbers. Using the wrong ones gets confusing. Its 25nm and 34nm I believe.
Dooms101's Avatar
Oh thanks for catching that! I think I said "35nm" in a different thread... let me go change that.
I.M.O.G.'s Avatar
I'm not sure actually how many different versions they have, but I haven't looked closer into it. Reading this storagereview thread it mentions 25nm and 34nm, but reading the actual review written by the same person, it mentions 25nm and 32nm.


I would expect OCZ has exact info listed on their site for anyone interested in confirming themselves.
dejo's Avatar
my thought is that they are still sticking to just performance based on atto, which is highly compressable data. asssd is non compressable data and I am sure is prolly still going to be gimped a bit from the 34nm nand.

what isnt addressed is if the p/e is affected by not giving up the extra 4gb of space to cover the 40% reduction in expected life cycle of the nand itself. also if 25nm nand is half the price of 34nm nand, and you paid the same price as said 34nm nand how could there have ever been a price upcharge to cover the price difference of the half priced ram? wow, is this ever a quagmire
jaymz9350's Avatar
I just picked up a 60GB V2 over the weekend unaware of this (bought due to the great praises of the V2).

Now like many I'm not at all happy about sneaking this in on us but the biggest thing to me for a manufacturer or seller is how they handle issues. This makes me much happier.
Dooms101's Avatar
Well fortunetly, OCZ has confirmed that they will not be making 8 IC configurations anymore and only the higher performing and larger formatted 16 IC versions. Just to clear this up again since tons of people seem to be confused about this, the 24nm IC's are actually better than the 34nm IC's in that they use less power, are cheaper (now anyways), and run cooler. It is the fact that OCZ decided to go to a 8 IC design when they switched to 24nm IC's that hindered performance. It's cheaper to make a 8x 8GB version rather than a 16x 4GB version. Thankfully, you can upgrade for no cost at all to the 16x 4GB 24nm version through OCZ.
gotFrosty's Avatar
Thank you for posting this article! I was one of the unfortunate ones to get one of the 25nm drives. I will be sending mine back to OCZ for sure. I'll update as soon as the process is complete if someone hasn't beat me to it.
Xaotic's Avatar
From reading their forums, the 16IC product is still scant on performance. They are looking to release a new firmware soon to aid in addressing the issue, according to their forum staff. On what level of improvement, he stated that "how much I do not know but we are trying." This is as close to a tacit admission of performance issues as I have seen from their staff or management related to the NAND or controller implementation. We do know from the StorageReview article that they are using a different controller revision for the 8IC units, but have not seen the internals on the 16IC version.

Hopefully, this will resolve the issue. The worst case scenario would require a respin of either silicon or PCBA.
Techboy10's Avatar
Hey look at that. Guess which drive I purchased for my new computer....

Trying to decide if I should send it back or not. I haven't been dissatisfied with its performance, but just knowing that it should be faster is something that might bother me.

Maybe I'll send it back over spring break while I'm sending my 460 to EVGA to step up to a 560 Ti since I won't be able to use my computer anyway.

Just my luck...
deathman20's Avatar
Are you sure its a 25nm version?
Techboy10's Avatar
Well this is the drive I have: http://www.newegg.com/Product/Produc...SSD2-2VTXE120G

I ordered in on January 4th and it has the E in the model name so that means its a 25nm version right?
Dooms101's Avatar
Not necessarily. Not all "E" drive models are the 8 IC "crap" versions. Wait a few more days until OCZ releases their tool which tells you which version you have.
Psykoikonov's Avatar

Seems this is spilling into Vertex 3 as well. Sample benchmarks meaningless ?
ratbuddy's Avatar
The sample benches are just for hype and are very misleading.

Check out this quote from anand:

Later on in the article we see this gem:

Translation: OCZ sent out ringers designed only to blast through synthetics.
Dooms101's Avatar
Not a shocker. I have not had a good experience with OCZ yet and this is not the first time they've done shady business. I do enjoy my Vertex 2, but only because I didn't get screwed. It's not like OCZ is the only one selling great SF1200, 64GB based drives either, I could have bought a G.Skill, Corsair, or Kingston drive for roughly the same price, capacity, and performance. I really don't see any point in buying from them again.
bcsizemo's Avatar
Not to offend anyone but I've never been impressed with OCZ. I've never purchased anything from them, but I have dealt with the products first hand and for the price and the reputation they seem to have I don't get it. At one time I worked for a small computer company that carried OEM/OCZ/Mushkin memory. (I and several other people pushed hard for the Mushkin, which was only a small step up in timings over the OEM stuff and certainly behind the OCZ.) Either way though we never had a problem with Mushkin, but we were RMA'ing probably 10% of our OCZ stuff for failing in non over clocked/stock boards...

I know companies can change over time, but I never get that impression with them.
Leave a Comment