Sign in with
Sign up | Sign in

Lost Performance: Not Just A Figment Of Your Imagination

The OCZ Vertex 2 Conspiracy: Lost Space, Lost Speed?
By

There are a number of reasons why 25 nm asynchronous flash can be slower than 3x nm flash. These reasons don’t apply to synchronous flash—the stuff you’ll see on the upcoming Vertex 3 drives from OCZ.

The number one reason that performance drops is the need for more error correction on the 25 nm node. This is related to the fact that the smaller geometry sustains fewer program/erase cycles. Error correction imposes processing overhead, though. I have to assume that OCZ, on its 25 nm-based SSDs, has to bolster the error correction it’s doing, impacting performance. Basically, near-term the company is taking a hit so that 5000 P/E cycles out, when uncorrectable bit read errors become a concern, SandForce’s RAISE technology can cope with and fix them.

The conclusion here isn’t completely random. While working on my benchmark results, I shared some of what I was seeing with OCZ. I noticed that streaming performance in Iometer was consistent from the 34 nm to 25 nm drives. It was only once we started hacking around with the 4 KB random reads/writes and benchmark patterns that performance dropped. Well, that’s precisely where processing overhead would come into play with an architecture like SandForce’s, which tries to work as efficiently as possible by mashing data together and sending it sequentially.

The company promised to supply an experimental firmware, which is by no means ready for public consumption, but is designed to address my concerns. I don’t have specifics on what the firmware does other than alleviate processing overhead a bit. But my guess is that it either dialed back the aggressive ECC or optimized the new algorithm being used. The optimization in the first iteration is relatively minor, but OCZ claims there is a lot more it can do to narrow the performance gap we’re observing currently. Do I think it's a coincidence that SandForce's second-gen controller family is also getting more advanced ECC? Not at all.

Now, consider the alternative to what OCZ is doing. Another vendor building SandForce-based drives using 25 nm flash (this is going to include everyone soon, by the way; the shift is happening industry-wide) can maximize performance today without bolstering ECC—risking outright drive failures in the future. Of course, we can’t put that to the test until our 25 nm drive see more extraneous use. But it's a scary thought. Serve up better numbers today, cross your fingers, and hope failures don't become epidemic three years from now? I certainly hope not.

What I do have currently is a run-down of a true apples-to-apples comparison: two 120 GB Vertex 2s, each with 16 chips, representing the 34- and 25 nm-based NAND devices.

Display all 55 comments.
This thread is closed for comments
  • 3 Hide
    Anonymous , February 21, 2011 5:10 AM
    Mushkin.com have a letter posted on the website that they will not be going to 25nm as the drives life cycle is less than half that of a 3*nm drive. Also the performance is not there.
  • 1 Hide
    cangelini , February 21, 2011 5:25 AM
    LeekayMushkin.com have a letter posted on the website that they will not be going to 25nm as the drives life cycle is less than half that of a 3*nm drive. Also the performance is not there.


    Depending on supply of 34 nm NAND, that's probably not a sustainable position to take. IMFT isn't going to decide to shift back to 34 nm.

    At the risk of contradicting myself, Intel will be using 34 nm NAND on its next-gen 6 Gb/s Elmcrest drives. It's not like the flash isn't disappearing entirely, but the vendors making the switch seem to be motivated largely by cost-cutting reasons.
  • 4 Hide
    radiovan , February 21, 2011 6:04 AM
    Nicely done Mr. Angelini; however, I still think OCZ pulled a nasty car-salesman tactic on their newest 25 nm NAND SSD products, and as such will be looking elsewhere to purchase any future SSD. Blaming resellers or other sources for the SKU is an incompetent way of deflecting fault and has made them look even more silly.
  • 1 Hide
    cangelini , February 21, 2011 6:17 AM
    Thanks radiovan. Like I mentioned in the story, we'll have to see if companies like Corsair and Patriot are able to get their upcoming renamed SKUs onto Newegg, Tigerdirect, Zipzoom, etc.
  • -1 Hide
    binoyski , February 21, 2011 6:40 AM
    Ok, but when will we be able to buy ssd's that are $100 @ 1TB capacity?
  • -1 Hide
    Anonymous , February 21, 2011 7:28 AM
    1.29 Firmware....how nice... pity that ocz only documents 1.28 for the moment.

    And then again why not update the 25nm to 1.29 too?

    But I guess consistency is not always wished when trying to demonstrate something
  • 1 Hide
    nebun , February 21, 2011 8:24 AM
    binoyskiOk, but when will we be able to buy ssd's that are $100 @ 1TB capacity?

    in about 10 years or so....that's a very big maybe
  • -1 Hide
    dconnors , February 21, 2011 8:29 AM
    Don't lie, Chris. We all know you still have (and love) that little Miata.
    -Devin
  • -1 Hide
    yose3 , February 21, 2011 8:29 AM
    binoyskiOk, but when will we be able to buy ssd's that are $100 @ 1TB capacity?


    when you throw a shoes to bush again meaby lol
  • 2 Hide
    Reynod , February 21, 2011 8:45 AM
    It is all about profit.

    The real reason is that going to cheaper flash with a 3000 cycle life to reduce the total cost of production (and therefore increase profit) means ... increasing the amount of redundant memory to replace the flash that dies due to wear ... and that process means better error correction is required to achieve that ... therefore performance is effected.

    Chris ... I got it into one sentence ... albeit a horrible one.

    Nice article mate.
  • 0 Hide
    cangelini , February 21, 2011 8:57 AM
    meanon1.29 Firmware....how nice... pity that ocz only documents 1.28 for the moment.And then again why not update the 25nm to 1.29 too?But I guess consistency is not always wished when trying to demonstrate something


    Update your drive with the 1.28 toolbox--it'll show up as 1.29 :) 
  • 0 Hide
    hannibal , February 21, 2011 10:06 AM
    Well normally moving to smaller production technology has mean better performance. I am not sure if the new situation where newer technolygy is inferior compared to old took manufactures by surprice?
    But in anyway their own tests should have shown it...

  • 2 Hide
    mike2100 , February 21, 2011 11:35 AM
    Mazda did the same thing in 2004 with the RX-8. Originally advertised 255hp dropped to 238 and it offered to buy back cars.

    Very bad move on OCZ's part. Justified or not, many people who hear this news will not be buying an OCZ drive.
  • 2 Hide
    JohnnyLucky , February 21, 2011 11:43 AM
    Great article.

    It appears as if OCZ was the first to market which would normally mean press releases, announcements, lots of hoopla, and technical reviews. Instead, OCZ chose to remain silent until customers complained. Silence had to be a management decision. What did management know?
  • 1 Hide
    JohnnyLucky , February 21, 2011 11:51 AM
    BTW - Veteran posters at Tom's Hardware might remember what happened when OCZ acquired PC Power and Cooling. PCP&C had some of the best high quality power supplies available until OCZ acquired the company.
  • -1 Hide
    saint19 , February 21, 2011 11:51 AM
    The same was for Toyota with his brakes problems few years ago.

    OCZ is a good manufacturer, but at this point when they are out of the RAM business, they need fix that problems and avoid futures ones if OCZ as company wants be on the market for more years in future.
  • 2 Hide
    rantoc , February 21, 2011 12:20 PM
    So they pushed out a cheaper less performing model with same name and stayed quiet about it, not the best way to get customers to return to the brand!

    I don't like the idea of 25nm flash for several reasons, reliability is my main concern. Sure there are better block handeling in the last gen sandforce controllers (vertex3) but what about quicker dying cells and when the controller runs out of spare cells! When it comes to storage i rather let the tech mature at the expense of early adopters (paying betatesters!?) than have to restore an old backup and hassle with RMA.
  • 1 Hide
    Anonymous , February 21, 2011 12:34 PM
    The "mistake" by OCZ was not a mistake at all.
    They are doing the same thing as before...
    With their Rally 2 USB drives to be specific.
    The Rally 2 did not have a model change at all (although it did have a different LED color), but users suddenly found out the new Rally 2 drives did not work in Xbox 360s and were slower.

    OCZ has been great with their SSD customer support, but this fiasco was a huge slap in the face to all of their fans.
    I know my next SSD will most likely be another brand unless the price is considerably less.
    It will most likely be a Corsair, Crucial, or Intel. Corsair, because they were completely up front with their transition to 25nm and will also continue to have the Sandforce based chips for their future SSDs.
  • 1 Hide
    compton , February 21, 2011 12:57 PM
    Good job, Chris.

    I'm glad the crew at Toms is on the case. A thorough analysis indeed. Keep it up.

    As a side note, I think I may stick with the 34u NAND drives for my third SSD purchase. I would always like more performance, but an SSD is basically destroying itself slowly. The biggest chunk of performance is gained simply by transitioning from HDD to SSD for OS and applications. Everyone wants a faster drive, but I also want my drives to last for a while. I'm concerned that reliability of ALL devices may suffer with smaller lithography. Maybe is't unwarranted concern. Time will tell.
  • 0 Hide
    scook9 , February 21, 2011 1:30 PM
    I got the 120GB Vertex 2 on the Jan 31 Shell Shocker on Newegg, since then I have been UNHAPPY with its performance. Now I know why.....I also had noticed the performance was lower than it should be

    I got the drive and threw it in the desktop to test before putting it in the laptop. Everything looked good. Threw it in the laptop and installed Windows 7 and a few basic programs. Performance now SUCKED (read was like 5% slow - 25% in 4k, and writes were all about %50 of what they should have been). Angry, I thought, maybe the laptop is bottlenecking it, threw the drive back in the desktop. Lo and behold, same abysmal performance. This was all with the drive having an on-time of about 45 hours since manufacturing......virtually NO use. I have now secure-erased the drive and that seems to have temporarily restored performance although the 4k speeds are still about 10 MB/s low (high 20's instead of high 30's)

    I have 6 SSDs right now and this is the only one I do not like. I am actually the happiest with the A-Data S599 rebrands that MicroCenter is selling (the $110 64GB ones - got 3).

    SO, I am interested in this thing that OCZ is doing, is that simply changing the chip density or also getting you a drive with 34nm chips on it? Will either do that or have newegg swap the drive with a different brand for me
Display more comments