Comments Locked

62 Comments

Back to Article

  • jrocks84 - Wednesday, January 5, 2011 - link

    Is the 25nm NAND from IMFT going to be DDR NAND?
  • Chinoman - Wednesday, January 5, 2011 - link

    Wouldn't anyone else like to see RAID 0 with TRIM support?
  • CharonPDX - Thursday, January 6, 2011 - link

    If the drive has good garbage collection in its own firmware, it doesn't need TRIM support.
  • crimson117 - Thursday, January 6, 2011 - link

    Garbage Collection is better than nothing, but never quite as good as TRIM.
  • kb9fcc - Thursday, January 6, 2011 - link

    Not needing either would be best.
  • Taft12 - Thursday, January 6, 2011 - link

    Citation?

    I'd be especially interested in seeing your statement backed up with regards to this hardware that isn't even released yet!
  • sprockkets - Thursday, January 6, 2011 - link

    Garbage collection works better when it knows what garbage to collect, hence TRIM.
  • GullLars - Saturday, January 8, 2011 - link

    Garbage Collection and TRIM are not mutualy exclusive, and are both ways of maintaining write performance. If you don't have any GC at all working with write placement algorithms, you end up with something like early JMicron devices.

    TRIM simply allows the drive to not have false-valid data, meaning it can mark blocks as dirty and have them included in the GC instead of moved around.
    This can significantly lower write amplification, and is more important the less unpartitioned space you have, as it allowes a second pool of dynamic freespace for GC to work with.

    BTW, i'm running 4R0 C300 and get about 1150MB/s read on SB850 with ATTO, and about 100k random IOPS. I don't think Z-drive R3 can beat that, compressible data or not.
    I also found 4x was overkill, 3x will get you past 1GB/s and 100k IOPS.
  • glugglug - Thursday, January 6, 2011 - link

    Garbage collection and TRIM are totally unrelated. No matter how good your garbage collection is, it absolutely does not overlap with or remove the benefit of TRIM.
  • bcronce - Thursday, January 13, 2011 - link

    I wouldn't say "totally unrelated". TRIM augments the GC.

    TRIM does nothing more than flag blocks for the GC to clean up which reduces the amount of "guessing" the GC has to do.

    Less guessing = better
  • goozira - Wednesday, January 5, 2011 - link

    With the IBIS XL 5.25" drive, what connection ports are on the back and is the box completely enclosed? I'm wondering if OCZ made the unit upgradeable with slide-out PCB slots or would you have to buy different levels of storage like the Z-Drive.
  • ckryan - Thursday, January 6, 2011 - link

    I wonder if the next gen SF drives will be the next step in performance. Moving from an Agility or Vertex to a Agility 2/Vertex 2 isn't that big a step... certainly not the giant leap moving from a mechanical drive to a SSD was. I wonder if moving from a 2g Intel or Idilinx drive to a nextgen SF drive will seem like that large of a leap in tangible performance. Getting that kind of performance leap is addictive, and why I replaced the drives in my desktop and laptop with SSDs to begin with. Upgrading from SSD to a marginally faster SSD doesn't make much sense, but it certainly seems like the next generation of SF's may have the leap in performance to make the upgrade worth while.
  • cactusdog - Thursday, January 6, 2011 - link

    No doubt these are going to be much more expensive too but unless you move around a lot of 4GB+ files it probably wont be very noticeable with normal computer use over a fast sata 2 ssd.
  • vol7ron - Thursday, January 6, 2011 - link

    - "No doubt these are going to be much more expensive too"

    Maybe market prices. The NAND should be cheaper, it's the controller that's driving the price. I'm not sure if the SF-2K controllers will be all that much more than the 1500s were when they were released. If what the article posted is true, this is the best controller, so it certainly will be expensive. I'm curious how the lower (cheaper) SF-2K controllers fair in performance.

    - "it probably wont be very noticeable with normal computer use over a fast sata 2 ssd."

    Maybe, maybe not. I still notice a slowdown with Intel G2 during startup, just because of the massive amounts of startup apps and widgets that now exist.

    With more screen real estate (those that have large/multiple monitors), you have more space to want to fill, and thus a larger demand on your hardware resources.

    Though, I still have my anti-virus/games on a mechanical drive to increase the longevity of the SSD. I'm waiting for that $/GiB ratio to go down before thinking about putting them on SSDs.
  • Out of Box Experience - Saturday, January 8, 2011 - link

    Can a single Vertex 3 copy and paste 4GB files faster than a modern 7200 RPM Western Digital?

    This could be interesting..

    A modern Western Digital desktop drive is 3 times faster than a Vertex 2 at copying and pasting data but a vertex 3 isnt 3 X faster than a Vertex 2 so its still probably slower than a Western Digital at copy/paste functions..

    You know, REALWORLD functions!
  • MrSpadge - Thursday, January 6, 2011 - link

    The faster storage becomes, the more of a bottleneck the rest becomes. If your CPU is already at 100% there's only so much an even faster SSD can do. So don't expect another huge jump, but rather solid improvements.

    MrS
  • RussianSensation - Thursday, January 6, 2011 - link

    Looks like SATA 3.0 is going to be obsolete by the time Vertex 4 rolls around. Unbelievable. I might just have to bite the bullet and finally grab an SSD drive, which of course means I have to dump my 1.5 year old S1156 platform with its by now archaic SATA 2.0 interface !
  • probedb - Thursday, January 6, 2011 - link

    Just buy a SATA 3.0 PCIe card :)
  • therealnickdanger - Thursday, January 6, 2011 - link

    Or avoid the whole mess and buy a Revo and get superior IO.
  • Shadowmaster625 - Thursday, January 6, 2011 - link

    Can you boot off a SATA 3 PCIe card? I mean without jumping through 17 hoops.
  • juhatus - Thursday, January 6, 2011 - link

    of course you can.

    Requires you to goto all the way to bios thou.. think server's all booting from the of pcie-cards as long as there's been pcie.
  • vol7ron - Thursday, January 6, 2011 - link

    Let's stop using the word/acronym "BIOS" and start using the term "UEFI" :)
  • UNHchabo - Thursday, January 6, 2011 - link

    There's a problem with that though: phonetics.

    It's easy to say BIOS: buy-ose. How are you supposed to pronounce "UEFI"?

    I suppose "you-fee" may work, but that "E" complicates things. Maybe we could try ü-fee?
  • Spivonious - Thursday, January 6, 2011 - link

    Ha, I like üfee. Best spoken with an Ahnold voice.
  • Taft12 - Thursday, January 6, 2011 - link

    Not a very good option on S1156. Fail chipset from Intel FTL :(
  • IanWorthington - Thursday, January 6, 2011 - link

    I assume that 1GB/s is sequential i/o? Would love to see the 4kb random speed.
  • ajp_anton - Thursday, January 6, 2011 - link

    It's all there in the graph. 4kB is about 200MB/s both directions.
  • ZippoMan - Thursday, January 6, 2011 - link

    Really? Q2? Damn. It will be interesting to see the C400/M4 face off with the Vertex 3.
  • cactusdog - Thursday, January 6, 2011 - link

    The C400 is crapola. Comes no where near the vertex3
  • vol7ron - Thursday, January 6, 2011 - link

    Remember the V3 is using an enterprise SSD controller. More expensioso
  • DoktorSleepless - Thursday, January 6, 2011 - link

    "50-400gb usable"
    Well, that's really disappointing. I was hoping for the 60-480 scheme from the current drives.
  • vol7ron - Thursday, January 6, 2011 - link

    I think you touched on something I wanted to bring up.

    I care less about how much manufacturers are using to make their drives work/perform and more about what can actually be used. The 50-400 is the important number and the one I urge companies to start using, uniformly. The other number should just be left in the specs for enthusiasts consideration.
  • strikeback03 - Thursday, January 6, 2011 - link

    Maybe with 25nm NAND they will go back to more spare area, but I wouldn't be surprised if the first drives out are more enterprise focused and have more spare area, like the SF-1200/1500 launch. Then maybe drives with less spare area would come later.
  • iwodo - Thursday, January 6, 2011 - link

    Once again we have Raw performance numbers.

    Tell me how an Toshiba SSD with lower Seq / Random Read Write numbers manage to beat an Sandforce or Intel SSD then we will start talking.

    Impressive as they may sound, i would like to see some real Anand benchmark before i make the judgement.
  • probedb - Thursday, January 6, 2011 - link

    Did you miss the "very, *very* early silicon and hardware"?
  • BugblatterIII - Thursday, January 6, 2011 - link

    We've seen before that the larger drives have far better performance due to their greater number of banks. Was it the 512GB drive you tested?
  • semo - Thursday, January 6, 2011 - link

    Interested about that too. The C400 for example seems still a mediocre SSD by today's standards at 64GB.

    "both Micron and Intel beating it to the punch" -- We know that the Micron/Crucials are coming out soon from your previous article but not sure what's the latest on the Intels

    "OCZ also demonstrated a new, slimmer chassis for its 3.5” SSDs like the Vertex 2 and Agility 2" -- Did you mean 2.5"? I guess they would be 7mm high?

    "Obviously to hit these speeds you obviously need a 6Gbps controller" -- obviously :)
  • jonup - Thursday, January 6, 2011 - link

    No, Anand meant 3.5" SSDs. These have been available for a while now. They are slimmer than 3.5" but fit in the normal 3.5" holes. I have the suspicion that you can fit two of these in a single 3.5" bay because of how thin they are, but I have not tested it since I have played around with only one.
  • semo - Thursday, January 6, 2011 - link

    Thanks didn't know that. I thought only the IBIS and colossus were available in 3.5"
  • vol7ron - Thursday, January 6, 2011 - link

    "I have the suspicion that you can fit two of these in a single 3.5" bay because of how thin they are"

    I'm sure you could, but are there any heat concerns?
  • evilspoons - Thursday, January 6, 2011 - link

    The Vertex 2 / Agility 2 dissipate like ONE WATT while being tortured (http://www.anandtech.com/show/3681/oczs-vertex-2-s...

    I hardly think it's a problem.
  • ppokorny - Thursday, January 6, 2011 - link

    I recently had a chance to use those new 3.5" drives, and they mix plastic and metal mounting holes in a way that makes the mechanical fit not be flush on the sides.

    That caused problems when I had to mount the drives in a hot-swap drive tray of a server.
  • marraco - Thursday, January 6, 2011 - link

    Of course, the Vertex 3 should saturate SATA 2 on sequential speed, but I wish to know how much slower performs on 4 Kb tests when connected to SATA 2.
  • Marc HFR - Thursday, January 6, 2011 - link

    Dear Anand,

    4K Random read is more than 3 times faster than SF-1200 one.

    Are you sure you 8GB LBA space restriction on this test ?
  • H8ff0000 - Thursday, January 6, 2011 - link

    A lot of us want to know this. I know there's probably no info on this, but some sort of ballpark figures would help OCZ's business. If people heard a ballpark figure that didn't scare the piss out of them, they very well might actually wait for this to come to market instead of buying a C400. If I remember correctly, they said somewhere around $200 for a 120GB C400. If this were $250ish ballpark I'd wait.
  • strikeback03 - Thursday, January 6, 2011 - link

    Problem is that if they threw out a number now for a product they are hoping to launch in 4-5 months, even saying it is "ballpark" people will complain if they don't hit it. Say they said $250 right now, and you wait, and they launch in June at $300 for 120GB. Some users would be extremely upset by the wait and the "price increase".
  • tjoynt - Thursday, January 6, 2011 - link

    Why the heck did Intel decide to name their "some % random" IOMeter write strategy "pseudo random"? "pseudo random" already means something and it is not that. :(
  • vol7ron - Thursday, January 6, 2011 - link

    Page 1:
    "I get 518MB/s sequential write speed and nearly 500MB/s for sequential read"
    - Are you sure that's accurate, or is it natural to have sequential write faster than sequential read?

    Page 2:
    '5.25” drive by'
    - Probably meant drive bay
  • glugglug - Thursday, January 6, 2011 - link

    On normally HDDs, random write is almost always faster than random read because the buffer cache used by the controller has a far more significant impact on writes.

    On an SSD maybe this applies to even sequential I/O as the wear leveling algorithm/ logical-to-physical block mapping may make your sequential operations not really 100% sequential.
  • FilipK959 - Thursday, January 6, 2011 - link

    Well, my old Pentium 3 PC the PC 100 sdram had a transfer rate 420Mb per second according to SiSoft Sandra if I remember correctly so this is just amazing that a mass storage device can pull this off. :))
  • B3an - Thursday, January 6, 2011 - link

    Can anyone tell me how i could RAID these drives with 6Gb/s on a motherboard with no SATA 6Gb/s?
    Are there any PCI-E RAID controllers for this that dont cost a ton?
  • tipoo - Thursday, January 6, 2011 - link

    When a startup can compete with a giant. Right now it looks as if Sandforce, with limited funds and only a relative handful of employees, is the only company that can put up a fight against the mighty Intel controller.
  • bobbozzo - Thursday, January 6, 2011 - link

    Please explain the compression comments and tables... is the drive compressing all data?

    Thanks,
    Bob
  • dertechie - Thursday, January 6, 2011 - link

    That's part of the Sandforce secret sauce. The control ASIC has some amount of (probably fixed function) compression/decompression hardware. This is what gets them their low write amplification and their fast speeds. As you can see, the controller is still fast if it can't compress the data, but not as fast as it is when you feed it something highly compressible.
  • janon123 - Thursday, January 6, 2011 - link

    What is the total 4KB random writes that can be done to each drive type in its lifespan?

    Intel X25-E (G3) 25nm eMLC 100GB is rated at 900TB-1PB total 4KB random writes.
  • janon123 - Thursday, January 6, 2011 - link

    and are any of these Vertex 3 drives capacitor-backed?
  • DarkRogue - Thursday, January 6, 2011 - link

    I would assume the slide that reads "Power loss data protection" would mean it includes a cap to flush data to NAND in the event of a power loss.
  • AbRASiON - Friday, January 7, 2011 - link

    Over 180mb/s 4k READS is what I've been waiting for.
    This SSD, coupled with an absoloute meaty CPU and lots of ram will be awesome.
    Forget sequential read and writes, they are mostly meaningless rubbish once they get over 200mb/s - this baby is going to load games and applications exceedingly quick. Damn glad I skipped the C300.

    I am eager I tell you.
  • qax - Saturday, January 8, 2011 - link

    I would like to see 2 of these controllers on af PCI-E PCB.
  • amjustice - Friday, January 14, 2011 - link

    I was holding off on getting a new SSD to put in my Macbook Pro until this next gen of drives comes out. Since I do not have the proper controller on my MBP to do 6GBPS is it worth it to wait still? Will I be able to see some of the benefits of speed in these drives?
  • grandnexus - Wednesday, January 19, 2011 - link

    So I'm reading about this Sandforce SF-2000 controller for SSD hardrives, and it's apparently the hottest shit ever. I'm planning on doing a rebuild soon, within a week or two and I was going to get an SSD hardrive.

    After reading this, it sounds like I should wait until this comes out. So, what is the release date of SF-2000? And are there any other comparable SSD hardrives that are out or will be out in a few weeks?
  • grdh20 - Friday, February 4, 2011 - link

    I heard from OCZ earliest end of this quarter or earl 2nd quarter. I say worth the wait.

Log in

Don't have an account? Sign up now