Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Intel Technology

No Intel Turbo Memory for Desktops Until Next Year 75

Might E Mouse writes "While Intel's 3-series chipsets support Robson/Turbo Memory, the general consensus amongst motherboard manufacturers at Computex is that we're not going to see the technology on the desktop until next year at the earliest. Working modules are on display at the show, but they're not going to be available to buy for a while."
This discussion has been archived. No new comments can be posted.

No Intel Turbo Memory for Desktops Until Next Year

Comments Filter:
  • by tknd ( 979052 ) on Thursday June 07, 2007 @09:31PM (#19432435)
    So does this mean they found a new place for the turbo button?
  • Turbo Memory is... (Score:5, Informative)

    by phantomcircuit ( 938963 ) on Thursday June 07, 2007 @09:40PM (#19432509) Homepage

    Intel Turbo Memory lets your notebook actually learn your habits to provide better system response. That's because it stores frequently used information near the processor, where it's more quickly available. Better CPUs run better with Intel Turbo Memory.


    This entirely new system innovation for Windows Vista PCs is based on Performance Intel® NAND Flash Memory (like the memory in an iPod* or USB 'thumb' drive), together with supporting software. It works alongside your system's RAM to increase the efficiency of data movement between the processor and hard disk.



    http://www.intel.com/design/flash/nand/turbomemory /index.htm [slashdot.org]">Intel® Turbo Memory

    • by king-manic ( 409855 ) on Thursday June 07, 2007 @09:49PM (#19432591)
      Intel Turbo Memory lets your notebook actually learn your habits to provide better system response. That's because it stores frequently used information near the processor, where it's more quickly available. Better CPUs run better with Intel Turbo Memory.

      This entirely new system innovation for Windows Vista PCs is based on Performance Intel® NAND Flash Memory (like the memory in an iPod* or USB 'thumb' drive), together with supporting software. It works alongside your system's RAM to increase the efficiency of data movement between the processor and hard disk.

      http://www.intel.com/design/flash/nand/turbomemory [intel.com] /index.htm">Intel® Turbo Memory


      Sounds like slow off chip cache, a la certain L3 Cache made of flash memory. I wonder what makes it notable? Size? cost? speed? Does it really help anything? It seems a large enough main ram would invalidate this or even the mere presence of on chip cache.

      • It would appear that Turbo Memory helps with reducing the time to access hard disk data.

        This decreases the need for hard-disk accesses, saving both time and power, which equates to increased performance and energy efficiency.

        This seems like it would be just as useful as increasing system performance. The advantages being ease of installation and cost. The disadvantages being lower performance than standard system RAM.

        Memory caching at the disk level seems like a much more promising technology

        • by moro_666 ( 414422 ) <kulminaator@gCOB ... m minus language> on Friday June 08, 2007 @03:26AM (#19434585) Homepage
          I don't know which miracle NAND pieces intel has invented this time, but if they still rely on the old reliable technology, it's slow. Yes sure it's faster than spinning around the disk, but i'll just prefer a 64bit amd box stacked up with as much ram as i can get, linux will take care of buffering the fs into there. Regarding the volatile thing, my machine hasn't crashed for a year, and if it crashes really bad, it can even screw up the data in the NAND, so it can't always help me, now can it ?

            If you're familiar how electronics works, you obviously understand that really big writes and reads that pass through this unit, will actually slow down, not speed up. You will have the latency of the card in addition to the latency of your hdd.

            One more card in your machine is one more thing that can break, and people who have used the NAND units (available so far) a bit more know that at some point they will start to break down. It has been written in their specs too, read those if in doubt.

            Perhaps the NAND trick is a great idea for laptops that suspend/resume a lot, but for other machines, just buy the ram. It's probably cheaper, it's faster, it's not locked down behind the iron curtains of Intel, and most importantly, it's already there.
          • by pipatron ( 966506 ) <pipatron@gmail.com> on Friday June 08, 2007 @03:46AM (#19434697) Homepage

            Perhaps the NAND trick is a great idea for laptops that suspend/resume a lot

            Yes. Laptops sell more than desktops these days. Apparently it also does a lot for the time it takes to boot.

            just buy the ram. It's probably cheaper

            Go and compare prices of 2GB DDR2 and a 2GB Compact Flash, you'll notice there's quite a big difference.

            It's not locked down behind the iron curtains of Intel

            Intel, which is one of the few hardware companies that actively work for open drivers, when the rest are trying as hard as they can to lock everything up.

            and most importantly, it's already there

            Yesterday I was browsing for a new laptop (well, I only bother with thinkpads), and guess what? Turbo Memory.

          • Re: (Score:3, Interesting)

            by Nikker ( 749551 )
            The benefits of using the same technology off of a removable device are much better aren't they?

            1. I can use it with my existing system
            2. The USB port is fairly well documented and addopted by current OS's
            3. Your point does seem valid about slowing down real time activity but good for suspends on laptops, so using this with a removable device would add security to your setup
            4. Portable reads & writes have more potential as people use different machines
        • Except for those who already have disks, and want to keep using them. This just means that Intel doesn't want to steal the thunder from the drive manufacturers already selling flash-added drives. It's not aimed at us here at slashdot(we had ours about how the technology some time ago), it's aimed at shareholders...
        • by monsted ( 6709 )

          Memory caching at the disk level seems like a much more promising technology

          Except this is disk caching and needs to be solid state to be terribly useful. True, memory caching is good enough for reads, but to really push write caching you need a solid state cache so your file system doesn't die when the power gets knocked out by lightning. If you put write cache in memory (as many current hard drives can be told to do), you'd better be absolutely sure that they don't get turned off by accident.

          Some storage systems from the big guys already use stricly controlled volatile write cac

      • by Vellmont ( 569020 ) on Thursday June 07, 2007 @10:25PM (#19432883) Homepage

        I wonder what makes it notable? Size? cost? speed?

        High throughput, low latency. Don't know about cost.

        It's basically the perfect hibernate cache that doesn't require power to maintain it's state, and will give near instant uptimes. You could also gain a bit from caching disk reads.

        It seems a large enough main ram would invalidate this or even the mere presence of on chip cache.

        RAM is volatile unless you constantly supply power. Because of this you can't rely on the information to still be their when you come back to a full power state.

        Basically this little device would allow people to turn off their PC completely, and power it back up into a fully functional state. You can sort of do that now, but it means either maintaining a little power to the memory to maintain state, or spending an interminable time writing out to disk.

        Of course, that means that driver writers need to actually support resuming from sleep, which many today don't properly support.
        • by cnettel ( 836611 )
          Another aspect is write caching. Imagine an idle machine churning out tiny registry writes (ok, blame it on the OS), or a moderately used http or db server where you still want complete logging and fs journaling. With this technology, the HD could spin down, and you would still be able to write to the filesystem and know that the things you put there, stay there. It touches on the hibernate issue, but I think the promise of possibly turning of the single remaining spindle in my machine is quite interesting.
          • If you used dual-ported Flash[1], then this could also provide greater throughput. If you buffered up a few GBs of writes before committing them to the disk, then you'd need a lot fewer head movements, which are the real performance killers for hard disks.


            [1] Note: May not exist, I haven't kept up with Flash much recently.

        • by fgouget ( 925644 )


          It's basically the perfect hibernate cache that doesn't require power to maintain it's state, and will give near instant uptimes.
          Oh, so now Windows users are complaining that Windows uptimes are too long. They would rather have their computer crash instantly as it comes up. Weird. I will really never understand Windows users. Better stick to my 3months+ Linux uptimes.
          ;-)
        • "RAM is volatile unless you constantly supply power." Yea, that's why we've had NVRAM for decades (Magnetic core memory, anybody?) Now it's going to PCRAM, which is perfect for storage but sucky for massive amounts of read-writes (even though trillions of read/writes is nice for storage, a normal game blows through that in a day or so, making it impractical for just general processing as a form of standard memory.)
      • by Slaimus ( 697294 )
        NAND? I always thought Intel only makes NOR memory.
        • It's a recent development. The NOR market is not growing as fast as NAND, and cellphone makers are starting to use more NAND thank NOR in new products. Intel is taking direct aim at Samsung, who is the NAND market leader.

          AMD is also working on a "hybrid" ORNAND flash with NAND interface and NOR device complexity, but it's not that impressive.
    • So where does the turbine fit in?
    • I don't doubt that this is the next type of cache which will make things faster, but you just have to love the way they make it sound like something new: "actually learn your habits to provide better system response!"
    • So, turbo memory is on-board flash memory for Ready Boost?
    • by Khyber ( 864651 )
      "That's because it stores frequently used information"

      Oh, you mean like Windows XP already does with my most commonly used programs so that after I quit them and run a couple other things and then return to the other programs they have faster loading times?
  • by lippyjka ( 1081173 ) on Thursday June 07, 2007 @09:52PM (#19432619)
    What the hell?; The inquirer posted an article about how MSI is going to bundle MSI to bundle Intel Robson cards with motherboards: http://theinquirer.org/default.aspx?article=40178 [theinquirer.org]. Who to believe? Bit-tech.net or TheInquirer.org..... I'm personally going with the inquirer...
    • Re: (Score:3, Insightful)

      by CajunArson ( 465943 )
      I don't know if I'd believe either source fully, but bear in mind the Inq's article is a photo from Computex, where lots of products that are not officially for sale yet are being shown. If I took everything I saw at Computex as something I could get immediately, then I'd be busy comparing Barcelona vs. Penryn servers with DDR3 RAM, and that ain't happenin' for at least 4 months.
    • It would be pretty funny if MSI and all the other Taiwanese manufacturers all agreed at IDF that they wouldn't deploy Robson on the desktop for another year.

      And then a bit later HULK SMASH! MSI start shipping Robson cache modules bundled with motherboards. That's the reason all the boards shipped with that funny connector on them marked "Factory test only, not a Robson connector".
  • by eebra82 ( 907996 ) on Thursday June 07, 2007 @09:54PM (#19432655) Homepage
    There is no reason for Intel to make this move just yet. Now that they are completely dominating AMD in the desktop segment, why not hold it as a safe card against AMD next time they come with something new?

    Intel would gain almost nothing on claiming another performance victory over AMD since it is widely known that their Core 2 Duo/Quad CPU series outperform AMD by a lot. So by releasing more technology that increases performance by a very small margin is like for AMD to announce the 1% speed bump with AM2 over 939.

    From another view, I think it is interesting to see that the laptops receive cutting-edge technology ahead of the desktop market. Could this become a trend in the computer industry?
  • Mmm, so what? (Score:3, Informative)

    by Junior J. Junior III ( 192702 ) on Thursday June 07, 2007 @11:36PM (#19433391) Homepage
    From what I've heard, Turbo Memory won't be supported by Windows XP, only Vista. Like a lot of sane people, and even most government agencies, I won't even think about running Vista until sometime next year, when they release a Service Pack or two that unfucks a lot of Vista's inherent shittiness.

    Is there/will there be support for it in OS X or Linux? It'd be nice...
    • by Sycraft-fu ( 314770 ) on Friday June 08, 2007 @02:50AM (#19434417)
      As far as I can tell, Turbo Memory is nothing more than flash RAM on the PCIe bus. It isn't what does the caching, Vista is. Vista has a deal called "ReadyBoost" that caches data to flash. You don't need this stuff, any fast USB flash drive will work. The USB interface works fine since max transfer rate of flash is pretty low. It is used for its fast access times. Basically Vista does a whole ton of caching, and does it aggressively.

      XP has pretty basic caching, it just leaves stuff in RAM. So if you open a program and then close it, XP doesn't actually zero the RAM it leaves it there. Should you open it again before the RAM is allocated for other uses, it'll use that again. Ok, fair enough, but that only helps for repeat launches. Vista does a similar thing, but actually preemptively loads things in RAM. It bases this off of your usage and thus what it thinks it needs to load the fastest. However since RAM is limited as a secondary cache, it'll use flash memory. Not good for large things, since it is slower for sequential transfers than a disk, but great for caching the first part of things. It starts to load off of flash while the drive seeks, then switched to the drive.

      It looks like the Intel memory is nothing more than Flash dedicated for this purpose (looking at the laptop card on their page it is just a controller and 2 flash chips). Thus the OS itself will actually need to do the caching. Linux and/or OS-X could of course add this, but it's up to the OS maker, Intel is just providing the memory on which to do it.
      • Re: (Score:1, Interesting)

        by Anonymous Coward
        I've been trying to understand why caching to a USB flash drive - instead of a hard drive - is a performance win. USB's max transfer rate is 60 MB/s; in practice it's more like 30-40 MB/s. If I have a moderately-non-sucky HD, it has a minimum transfer rate of 30 MB/s. A good SATA drive can do 60-90 MB/s: (according to StorageReview [storagereview.com]).

        I suppose the USB drive does eliminate seek time and rotational latency, but it's not obvious to me that this is a performance win. Has anyone seen benchmarks?

        And yes, Turbo
        • That might be to true if you're reading large sequential (not fragmented) data.
          In real life, its more like: write hundreds of thousands of small chunks (few KB each) to many different locations.

          The seek time dominates the performance - that's why flash storage wins here.
      • Can't say for sure, but since this disk is 'internal' Vista might disable the encryption/decryption overhead for it. On external drives they need to encrypt the data to prevent loss of sensitive information.
        • Isn't there a saying that if you have physical access to the device you're already hosed? I guess they can hold themselves to higher standards, but if your laptop has been lost, it's not like I can't rip the chip out or soldier wires to it.
      • by dave420 ( 699308 )
        The turbo memory is a much better solution than ReadyBoost on USB, as this can be used to boot the OS, whereas Vista's ReadyBoost on another device can't.
      • pci-e has a lot less cpu over head then usb.

        also where are the firewire flash ram sticks.
      • by aka1nas ( 607950 )
        This can be used for hybernation purposes, where readyboost cannot. It's still rather useless for desktop users.
    • Re:Mmm, so what? (Score:4, Interesting)

      by Ant P. ( 974313 ) on Friday June 08, 2007 @05:03AM (#19435017)
      Linux already has a patch to support a "prefetch partition", all it'd take to support this is making it show up as a mountable device.
    • AFAIK, the Linux kernel already has the MTD (Memory Technology Devices) framework to support things like flash cache - all that's needed is an additional driver for the Robson/whatever flash tech, and presumably some sort of userspace app to do all the prefetch/caching etc. I imagine the same would be true of hybrid hard drives.

      A few people, myself included, already use MTD for using graphics card VRAM as swap (server doesn't have onboard GFX on it and never uses X, so I chucked in a dirt cheap nVidia card
  • Turbo Memory is a cooler than DDR2 but I prefer Super DDR2 Turbo Hyper-Fighting Edition... it's SO much better than DDR3.
    • by empaler ( 130732 )

      Turbo Memory is a cooler than DDR2 but I prefer Super DDR2 Turbo Hyper-Fighting Edition... it's SO much better than DDR3.
      I prefer Super DDR2 Alpha. Or plain vanilla Super DDR2. I'm weird that way, I know.
  • Imagine what they'd do if these companies started working together instead of planning each others failures.

    They've figured out the nand deterioration over time with bitspreading/badbit mapping then? ... enough to rely upon, I'd imagine. What about the speed issue of nand? How does the speed compare to DDR2? PCI-E 16x is a boatload of bandwidth, tho with resume getting better it might make more sense to put in a crapload of ram and a good vfs cache. Do any of the current fs's support tunable memory ca

  • by Anonymous Coward
    http://news.zdnet.com/2100-9584_22-6188522.html [zdnet.com]

    Apparently it's not all it's cracked up to be.

    http://www.anandtech.com/cpuchipsets/intel/showdoc .aspx?i=2985&p=4 [anandtech.com]

    Apparently just more marketing hype.
  • I just checked my in box again and it has not shipped yet... I ordered the new Santa Rosa [bit-tech.net] offered from Sony (pre-sale only) last month. I did not realize just quite how exclusive it was when I saw it. Then I shopped around and found no one else was offering it.

    Seems like a decent deal in a market where most products are nearly obsolete before you get them home. Having this chipset gives me the ability to upgrade to 4GB RAM if need be. From what I hear of Vista, I may just want to do so.

    • by empaler ( 130732 )

      Seems like a decent deal in a market where most products are nearly obsolete before you get them home. Having this chipset gives me the ability to upgrade to 4GB RAM if need be. From what I hear of Vista, I may just want to do so.

      From what I know of Vista, you'd probably be better of sticking with XP or Linux. Upgrade your memory anyway. I tried Vista for two months on my work computer, but reverted before my heart popped.
    • Again (see above), Sony will have Santa Rosa but will not be releasing notebooks with Robson/TurboMemory this summer. This has actually become a spat between Sony and Microsoft, becuase Sony is claiming that Vista will not support Robson until SP1! (see http://news.zdnet.co.uk/hardware/0,1000000091,3928 7431,00.htm [zdnet.co.uk]) What's interesting about this is the fact - according to Sony - that M$ left the support out of Vista deliberately, because it was under such pressure to get the OS out. I wonder what else they
      • Yeah, from the article I linked to it seemed like all of Santa Rosa would use the turbo memory. I've been reading a lot more since then, and have come to a glaringly obvious conclusion.

        Stop posting at 3am while inebrei inebr iinnebir ... drunk.

        Or at least finish reading up on the subject as there is no unsubmit button on /.

        -

        Sig? what's that and where can I get one?

  • It would be nice to have dedicated flash memory for the operating system of your choice. I have no idea how this would work with linux or OSX, etc etc, but somewhere to store page files or hell, even the entire OS (although I'm guessing that would take a bit of work on the OS side). I guess it would make the motherboards much more expensive, so hell, even a slot to put in an SD card or something would be pretty handy..

I have hardly ever known a mathematician who was capable of reasoning. -- Plato

Working...