If geeks love it, we’re on it

OCZ Vertex 3 review

OCZ Vertex 3 review

 

OCZ Vertex 3 review

Fast is good. Faster is better. When the OCZ Vertex 2 first came out, the world was still mostly using SATA II ports, which carried a maximum of 3Gb/s throughput, or about 300MB/s. Not coincidentally, this is almost exactly what the SandForce-based Vertex 2′s performance numbers were. Of course, with new chipsets from both AMD and Intel, SATA III became more common. The new standard allows up to 6Gb/s per port, or upwards of 500MB/s transfer speeds. Of course, what good are those kinds of speeds if you don’t have a device that can use them?

The Vertex 3 uses a second generation SandForce controller, the SF-2281 to be exact. That controller provides support for capacities up to 512GB (~480GB usable depending on overprovisioning levels) and speeds of up to 500MB/s (SATA III level). Of course, OCZ has also customized their firmware to provide an even greater level of performance from the controller, as you’ll see below.

Here’s a rundown of some key technologies in the SandForce 2200-series controller:

  • DuraWrite helps to extend the life of an SSD through wear leveling and compression. Current NAND flash memory has a life expectancy of about 5,000 program/erase (P/E) cycles. Using DuraWrite allows the SandForce controller to determine the best location to write data that would have the least impact to overall SSD lifetime. Controller-based compression means data is written to fewer storage blocks, further reducing the wear on memory cells and increasing write speeds.
  • Hardware support for AES 256-bit and 128-bit encryption is also built into the controller.  Data written to a SandForce 2200-based SSD is automatically stored using a double encryption method.  If you want to keep your data safe, a disk-level password can be implemented—the password is required prior to system boot to enable access to the drive.
  • TRIM support is also present in SF-2200 series controllers, and can be used in Windows 7, Linux (kernel versions 2.6.33 and later), and Mac OS X Lion.  For operating systems without TRIM support, garbage collection is also implemented at the controller level.
  • Overprovisioning also returns in the SF-2200 controllers. This is a withholding of a certain portion of total capacity for garbage collection, wear leveling, and mapping of bad blocks. In the review drives’ cases approximately 16GB of the actual 256GB is reserved, leaving 240GB visible to the user.

For the review, results from the Vertex 2, RevoDrive X2, and a 7200RPM hard drive are included. So how does the Vertex 3 stack up?

OCZ Vertex 3 OCZ Vertex 2 OCZ RevoDrive X2 WD Caviar Blue
Available capacities 60GB – 480GB 40 – 240GB 100 – 720GB 80GB – 1TB
Maximum read 550MB/s 285MB/s 740MB/s 126 MB/s
Maximum write 520MB/s 275MB/s 690MB/s not listed
Random 4K write 60,000 IOPS 50,000 IOPS 100,000 IOPS not listed
Power consumption 1.65W (idle)
3W (active)
W (idle)
W (active)
4.3W (idle)
8.3W (active)
9.25W (max)
Warranty 3 years 3 years 3 years 3 years
Price (Current on 30 Sep) $134.99 (60GB)
$209.99 (120GB)
$439.99 (240GB OEM)
$569.99 (240GB Retail)
$1199.00 (480GB)
$94.99 (40GB)
$99.99 (50GB)
$104.99 (60GB)
$134.99 (80GB)
$154.99 (90GB)
$165.99 (120GB)
$246.99 (180GB)
$314.99 (240GB)
$349.99 (100GB)
$447.99 (220GB)
$469.99 (160GB)
$639.99 (240GB)
$1099.99 (360GB)
$1349.99 (480GB)
$2449.00 (720GB)
$33.99 – $64.99

On paper, it looks like the Vertex 3 should be roughly twice as fast as the Vertex 2, moderately slower than the RevoDrive, and make the Western Digital hard drive look absolutely ancient.

Synthetic Benchmarks

Sandra Disk Suite

Sandra disk test for OCZ Vertex 3 SSDSandra disk latency test results for OCZ Vertex 3 SSD

Sandra’s physical disk testing shows the Vertex 3 is about twice as fast as the Vertex 2 and about 12-20% behind the RevoDrive X2 for raw speed and all three are virtually tied for latency. Of course, the hard drive is left in the dust on all counts.

There was a problem performing the write test on the Western Digital hard drive, hence the zero score for both speed and latency.

Atto

Atto read test for OCZ Vertex 3

Atto write test for OCZ Vertex 3

 

Atto shows the best case scenario for SSDs—the data is highly compressible and is sequentially read and written.  The Vertex 3 really shows off what a good SATA III controller can do— in this case it’s the controller in the Intel P67 chipset.  Under most of the data sizes, the Vertex 3 nearly doubles the performance of a Vertex 2 and even approaches RevoDrive X2 speeds.

AS-SSD

Just as Atto shows the best case for an SSD, AS-SSD shows the worst case scenario—the data is incompressible, slowing the SandForce controller down.  The Vertex 3 really shows a huge lead over the Vertex 2 on sequential reads and writes.  What’s even more impressive is the nearly universal lead it holds over the RevoDrive X2 (a quad Vertex 2 in RAID-0 SSD) in the write tests.

CrystalDiskMark

CrystalDiskMark tells the same story as AS-SSD: univerally much faster than the older Vertex 2, almost as fast as the RevoDrive X2 for reading, and significantly faster than the RevoDrive X2 for writing.

Real-world Benchmarks

PCMark Vantage HDD

 

The Vertex 3 is significantly faster than the Vertex 2 using PCMark Vantages test cases, and in a few cases it even matches or outperforms the RevoDrive X2.

Load Times

When it comes to getting your computer to a usable state, an SSD is the best way to go. In this case though, there just isn’t much difference between the Vertex 3 and its predecessor. One and a half seconds for booting and a margin-of-error lead in loading Crysis isn’t a whole lot. Still, it’s way faster than a hard drive.

RAMDisk Copy

The charts above are two different ways of looking at the same data. The first chart is the number of seconds it took to transfer a folder containing just over a gigabyte of files to and from a 4GB RAMDrive. The second is the transfer rate in megabytes per second. The Vertex 3 once again performs significantly faster than the Vertex 2, nearly doubling the performance.

There was an odd caching problem with the RevoDrive X2 that caused flawed results, so those have been omitted.

Price

The 240GB unit is currently available for $439.99 for the OEM version, placing it in enthusiast territory. Considering the fact that in some cases it can outperform the RevoDrive X2 it seems justified. Most folks will probably want to look at the smaller and less expensive models, which carry more modest premiums over their respective Vertex 2 counterparts.

Conclusion

Icrontic Stamp of Approval BronzeThe Vertex 3 certainly delivers the SATA III speeds. Thanks to the customized SandForce controller, it’s even able to exceed the specs advertised by SandForce. If you’re already running a SATA II SSD, especially a Vertex 2, there might be some benefit to an upgrade, but it really depends on the application. New system builders or folks upgrading from a mechanical drive as their primary drive will definitely want to give the Vertex 3 a look—especially since SATA III ports are pretty much the standard now.  The extra speed does justify the extra cost if you can fit it in your budget.

The Vertex 3 is the fastest MLC SSD we have ever tested. It is awarded the Icrontic Stamp of Approval.

The OCZ Vertex 3 OEM is available from Newegg in 60gb, 120gb, and 240gb capacities. The full retail version can be found at Buy.com and TigerDirect as well. Finally, they’re all available  from Amazon:

 

Comments

  1. Thrax
    Thrax Mmmm, Vertex 3.
  2. fatcat
    fatcat What motherboard and/or SATA3 controller did you use?
  3. ryan
    ryan My vertex 2 is lucky I like it, or I'd replace it due to this review lol.
  4. Tim
    Tim Believe it or not, I'm thinking of getting a Vertex 3 SSD for the boot drive in my next gaming PC. Speeds have gone up and prices have come down enough for me to consider them. I'm thinking a 128 GB model would be good for me, with a couple 1 TB drives for storage.

    The article mentions a P67 chipset, I wonder if some of the newer chipsets do better or worse with SSDs than others.
  5. Bandrik
    Bandrik Adding this to my wish list. Great review, Nick! :D
  6. mertesn
    mertesn
    fatcat said:
    What motherboard and/or SATA3 controller did you use?
    It was an ECS P67H2-A2 using the native Intel controller. I also have results from the AMD 890FX and AMD A75M systems. The numbers aren't in front of me, but I recall them being somewhat comparable.
  7. Filipe Oliveira
    Filipe Oliveira This review is about as unbiased and impartial as a PIG selling sausages.... Anyhow, who actually believes benchmarks represent the real live speeds, deserves to be ripped off.
  8. PirateNinja
    PirateNinja
    Filipe Oliveira said:
    This review is about as unbiased and impartial as a PIG selling sausages.... Anyhow, who actually believes benchmarks represent the real live speeds, deserves to be ripped off.
    I NO RITE!!!!!!!!!111
    No reputable tech sites use benchmarks to review things, well all know that benchmarking is a garbage concept. I really prefer personal opinions based on 1-10 scales.

    Also, no other sausage exists besides pork based sausage. Ever.
  9. GHoosdum
    GHoosdum
    Filipe Oliveira said:
    This review is about as unbiased and impartial as a PIG selling sausages....
    Was your analogy meant to be ironic? Why would a pig sell sausages? Wouldn't it be asking for death?
  10. primesuspect
    primesuspect Internet tough guy alert
  11. mertesn
    mertesn
    Filipe Oliveira said:
    This review is about as unbiased and impartial as a PIG selling sausages.... Anyhow, who actually believes benchmarks represent the real live speeds, deserves to be ripped off.
    Sure I used some synthetic benchmarks, but here's where you argument falls apart: Real usage cases were used. Windows boot times, Crysis level loading, copying files to/from the SSDs - all common tasks. Even PCMark Vantage's HDD suite uses real applications for testing.

    So where's your evidence that I'm misrepresenting anything?
  12. primesuspect
    primesuspect He has none. He's just another pissed off, mentally unstable Internet Tough Guy.
  13. Filipe Oliveira
    Filipe Oliveira I have enough proof, real time testing, is using reported figures by Operative systems, sure no doubt you gain boot times and access speeds improve but nothing as dramatic as benchmark products present as results.

    However it is my experience with Sandforce vertex 2 drives that the advertised speeds vs real time speeds is half of what the benchmarks show, especially as main OS drive, I have discussed this with OCZ in great lengths via email and provided them with proof, even using the intel AHCI driver, instead of 250MB/s you get at best 100Mb/s slowly declining to 70MB/s on sata2 controllers.

    @primesupect, please keep your derogatory, offensive remarks to yourself, verbal abuse? Really? Please!!! We all watched lethal weapon, your the internet coward, come around to my neighborhood and insult me to my face.
  14. storrm
    storrm still think im going to go for an m4
  15. Thrax
    Thrax So, if you're getting less than 100MB/s, you're actually doing it wrong. I know this must be hard for you to accept, being one of those people who obviously jump to ridiculous conclusions that everyone else knows is wrong, but that's the case. You had a bad drive (unlikely), or you simply fail at testing a hard drive.

    Either way, -1 point for being one of those Internet tough guys that tells everyone to say things to their face. What are you going to do? Beat him up? LOL. What a joke.
  16. BuddyJ
    BuddyJ Lol. Filipe, primesuspect is from Detroit. He'll gladly tell you to your face. He'll slap your mother. He'll shoot your dog. In the back. Without shedding a single tear. Because he's one mean mother. He's the baddest man in the whole damn town. Badder than old King Kong. Meaner than a junkyard dog.
  17. mertesn
    mertesn
    Filipe Oliveira said:
    I have enough proof, real time testing, is using reported figures by Operative systems, sure no doubt you gain boot times and access speeds improve but nothing as dramatic as benchmark products present as results.

    However it is my experience with Sandforce vertex 2 drives that the advertised speeds vs real time speeds is half of what the benchmarks show, especially as main OS drive, I have discussed this with OCZ in great lengths via email and provided them with proof, even using the intel AHCI driver, instead of 250MB/s you get at best 100Mb/s slowly declining to 70MB/s on sata2 controllers.
    I also own multiple Vertex 2 SSDs of various sizes. I have seen none of the performance issues you claim. How do I re-create the circumstances that cause your sub-par performance? I'd like to see if it really does work.
  18. Filipe Oliveira
    Filipe Oliveira @mertesn

    http://bit.ly/qXjAnd

    that a review on tomshardware...

    recreating I suppose its a question of using same motherboard same hdd same cpu and same ram same drivers same firmware etc (they all have some effect on SSD/HDD performance). a pastebin of the specs http://pastebin.com/UGhvBUd1. Its been confirmed that a degradation occurs very quickly.

    Now to put this at rest hybrid drives do perfom much more consistently and suffer no such degradation, according to tomshardware a new bios is on the works to address this.

    As a side suggestion comments should go under moderation before being shown, We all love freespeech not freesheep and all.

    keeping ontopic...
  19. duraz0rz
    duraz0rz Is the scale on the ATTO tests wrong? It looks like it should be Speed in bytes/s.

    Personally, I'd like to see some tests that use uncompressable data to really test these SandForce controllers.

    Filipe, your experience may dissuade you from OCZ's Vertex 2 drives, but others have had great experiences with them. You can't assume that one drive is representative of the whole.
  20. BuddyJ
    BuddyJ Filipe, if you want SSD reviews by people who understand their complexities, look to The Tech Report or Olin's work at Benchmark Reviews if you don't trust Icrontic's stance. THG's credibility is hit-and-miss at best.
  21. Filipe Oliveira
    Filipe Oliveira mertesn

    here's a test system

    http://pastebin.com/hQQQFHta

    Over 28 drives and all of same were used tested found faulty rma'ed total SSD's on this system 1 (OS drive) others are all hybrid or plain mechanical, outperforming by miles vertex2.

    On other Systems varied configurations, further issues of similar failures and degradation.

    Why is OCS vertex2 a warning IMHEO is regarding other professional opinions and reviews like http://bit.ly/qXjAnd @ TomsHardware who are IMO a knowledgeable trustworthy unbiased source of information, looking at the products as a whole and not as they perform fro the duration of testing.

    duraz0rz

    Yes incompressible data like i.e. .AVI etc are a good test, however its not just attos we looking at, according to OCZ (Eric Ryder)they use attos and AS SSD to publish the figures published on website, another bit of obscure info that I didn't know until I presented attos benchs

    Just received a rma form to send back yet aother drive which on benchmarks its all fine but in real life is not as advertised by a long shot, and this is from OCZ themselves, not from any online/internet reviewer etc.

    All I say is beware not all that glitters is gold.
  22. GHoosdum
    GHoosdum Filipe, it is encouraging that you're engaging in discussion rather than taking a non-member shot across the bow with as many folks tend to do on our site with comments like your first.

    However, you still haven't addressed my question in regard to your analogy. In my opionion, a pig would not sell sausages out of self interest; rather, it would likely behave like the cows in Chik-Fil-A advertising by attempting to sell the meat products produced from other animals, in order to ensure its own survival.
  23. Bandrik
    Bandrik Perhaps the pig is actually a spy. I'm sure it was smart enough to make a secret deal with the carnivores to persuade other pigs to the slaughter, while enjoying protection and luxurious kickbacks.
  24. primesuspect
    primesuspect Just got one today. I upgraded from a WD Raptor 10K that topped out at 56mb/s on ATTO.

    From that, to this:
    image

    I cannot describe how much faster my computer is. What the fuck.

    Buy one NOW.
  25. mertesn
    mertesn Agreed. On most systems now the storage is the only real bottleneck, and upgrading to an SSD makes all the difference in the world. I put the Vertex 3 in my laptop as the boot drive and use a Seagate Momentus XT as my data drive. I will never go back to a spinner as a primary boot device.

Howdy, Stranger!

You found the friendliest gaming & tech geeks around. Say hello!