1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Storage With no sign of SATA4 coming soon, where do SSD's go to improve speed?

Discussion in 'Hardware' started by Farting Bob, 29 Feb 2012.

  1. Farting Bob

    Farting Bob New Member

    Joined:
    21 Jan 2009
    Posts:
    469
    Likes Received:
    13
    I was just looking at the SATA wikipedia page and there isnt a single mention of what will follow SATA 3. I coudlnt find any concrete evidence that SATA 4 is anywhere beyond the early planning stages.
    From this i can take it that any follow up offering faster speeds will still be quite a few years away. But unlike previous generations of SATA drives, SSD's are already pushing up against the limits in top speeds, and can only presume that they will continue to get faster with each generation.

    So where do SSD makers go from here? For marketing and genuine performance reasons i doubt many SSD manufactures would be happy to stay in the 500-550MB/s range for years and years.

    PCIe is the obvious first choice, but it has disadvantages.
    1. Its more expensive to make as you need the board and the extra chips needed currently(particularly an issue for smaller capacity and lower end drives, not so much for the bigger ones)
    2. We are used to having anywhere from 4 to 8+ SATA ports on a motherboard taking up very little space, and the drives themselves sit in the otherwise empty area of the case.
    But PCIe takes up alot of space on the board, and generally most boards will only come with 3 or 4 slots, and if you have a double wide GPU that will often leave you with 1 or 2 left for everything else to plug in. MicroATX and smaller varients are even more effected by the lack of expansion. My ATX board has just a single PCIe slot free right now thanks to my GPU taking up 2 and my network card using one, so it would have to sit right under the GPU, which will block airflow and is far from ideal.

    I suppose you could use Thunderbolt, especially if it gets up to its supposed top speed of 10Gb/s, with fiber offering even higher speeds in the future. But since its an active cable it will always be expensive compared to SATA on that front, and it will be a long while before we see more than 1 or 2 TB ports on motherboards. Advantages though include power transmission and future development.

    PCIe SSD's have their place in enterprise already, but we have yet to see any suggestion that the consumer market is going to migrate that way, and it may always remain a very high end option for those willing to pay very high prices for absolute speed.

    Or will most consumer SSD's just be happy to sit on SATA 3 for years? Top speed wont change but at least that might encourage controller makers to improve random read/writes and IOPS instead (not that these are really lacking in current SSD's, but its an area where they could improve without being blocked by the interface).
     
  2. PocketDemon

    PocketDemon New Member

    Joined:
    3 Jul 2010
    Posts:
    2,107
    Likes Received:
    139
    Not at all - the next standard is SATA Express which, whilst it appears to be slightly behind schedule in being finalised, will give 8 & 16Gb/s speeds...

    Now, in that, as far as i can see quickly, SATA 3 was finalised in early-mid 2009 & higher end pcie cards were out within that year that supported it... ...this 'should' mean that there will also be higher end pcie cards supporting SATA Express this year.

    it's then just the delay in them entering the consumer end of the market - enter (probably) a shonky Marvell version & then proper intel & amd on board solutions...

    ...what there is this time though is a clear driving force for quicker adoption d.t. the SSD market, but exactly to what extent this translates into a faster shift...???


    As to SSD development, shorter term the big moves forward will be in low end speeds &, for Sandforce, further reducing the difference between highly compresses & highly compressible r/ws...

    ...the former is partially dependent upon the shift to 128Gb 20nm intel/micron nand (& the equivalent toggle nand from Toshiba & whoever else) with the increase in speeds from onfi2 (up to 200MB/s) to onfi3 (up to 400MB/s) which should be in the SSDs towards the tail end of this year...

    ...though, along with improvements in low end, the gains are likely to be seen more heavily in the reduction in cost per GB - not least as there's the 64Gb 20nm parts coming in first which are still onfi2... ...& also, on the budget end, the introduction of TLC parts which will dramatically lower cost - though the price is much lower longevity.


    Anyway, that's a brief run down of what's known to be in the pipeline...

    Okay, there's always the chance that something like OCZ's HDSL interface (which was released as an open standard &, on the pcie card side, was comparatively dirt cheap) could see a resurgence & be adopted instead since it already exists - that having a bandwidth of 20Gb/s - however there's been no info about any moves with that in a long time so i wouldn't be putting money on it.


    [Edit]

    Forgot to type a bit of the conclusion that was in my head, but it should have been something along the lines that -

    ...the general point is that, whilst waiting for SATA Express to be available, there's both the potential for an 'arms race' at the high end of the market with improving the lower end speeds (large sequentials, whilst important, are not the only metric) &, more importantly for the manufacturers, being able to vastly increase the customer base through lower prices via a couple of different methods.

    & then of course remember that, whilst the C300 wasn't the best drive in the world by any stretch (it was just the first consumer one with >3Gb/s speeds), using its release timing as a guide then we'd be potentially looking at the first SATA Express consumer drives being in ~April 2013 at the latest.
     
    Last edited: 29 Feb 2012
  3. GoodBytes

    GoodBytes How many wifi's does it have?

    Joined:
    20 Jan 2007
    Posts:
    12,300
    Likes Received:
    709
    Another place that SSD can improve speed is, read/write speed of small files, because you see, the marked speed is always "best case scenario", and that is with a very large file. And also reliability can be improved.
     
  4. Paradigm Shifter

    Paradigm Shifter de nihilo nihil fit

    Joined:
    10 May 2006
    Posts:
    2,076
    Likes Received:
    37
    I dunno; speeds are pretty good on the higher-end ones right now. I'd be more interested in reliability (write cycle) improvements, capacity improvements and prices decreasing. That's enough to be getting on with. ;)

    The write cycle thing is of most concern to me... it seems every time a 'new' NAND tech arrives, it comes at the cost of decreased lifespan of the NAND cells. I'd quite like the option of still having usable SSDs in 15 years, like a 17 year old HDD is still doing service for me in a hardware firewall.

    edit: Post not in response to Goodbytes - he's right, small file read/writes could be improved a lot.
     
  5. play_boy_2000

    play_boy_2000 It was funny when I was 12

    Joined:
    25 Mar 2004
    Posts:
    1,414
    Likes Received:
    48
    If SATA express isn't expected to tip up in the next ~18 months then it wouldn't suprise me to see some sort of controller that takes 2 SATA connections and RAIDs them together (to a single drive) somehow. It would be awesome if it worked similar to the way that PCIe lanes or LACP work, where it's treated as a single interface, without the need for special drivers to support TRIM.
     
  6. Bloody_Pete

    Bloody_Pete Technophile

    Joined:
    11 Aug 2008
    Posts:
    6,332
    Likes Received:
    236
    I've never seen a SHDD rated at 15 years, but most SSD's are. This R/W cycles stuff is utter rubbish...
     
  7. The_Beast

    The_Beast I like wood ಠ_ಠ

    Joined:
    21 Apr 2007
    Posts:
    7,379
    Likes Received:
    164
    I don't have a SSD but I'd rather have bigger/cheaper drives than a smaller/faster one.



    Can current SSD fully saturate a SATA 3 connection?
     
  8. Elton

    Elton Officially a Whisky Nerd

    Joined:
    23 Jan 2009
    Posts:
    8,575
    Likes Received:
    189
    Time for improvements in storage space. And improvments most of all. In density price.
     
  9. play_boy_2000

    play_boy_2000 It was funny when I was 12

    Joined:
    25 Mar 2004
    Posts:
    1,414
    Likes Received:
    48
    SATAIII tops out at 600MB/s after overhead, and the current gen controllers are right on the doorstep (450-550MB/s) in sustained read. There are other aspects that controllers can be improved in, but with those improvements will generally come better sustained read/write (and we can't have that go to waste, now can we?).

    Dammit, I suddenly feel the need to upgrade my SSD :blah:
     
  10. Guest-16

    Guest-16 Guest

    This is not a chipset issue though, but it is ultimately the driving factor.

    Chipset makers will look at the mainstream market and how to get the best out of that while still turning a profit (cost down designs, smaller FAB etc). At the high-end you'll be looking at PCIe 4x-8x cards come down in price but still be premium priced over the current crop of high-end models I expect. Other than that I would expect more improvements in ECC, write amplification and reliability - if anything this time is needed to hammer out the bugs for long term benefits of the solid state market.
     

Share This Page