There is a lot going on for Intel
Link to Engadget
Basically there are a few new series that will be available this year
710 MLC-HET
720 PCIe-SLC
520 MLC
20GB Larsen Creek SLC?
Attachment 114040
Attachment 114041
Will read up on this later...
Printable View
There is a lot going on for Intel
Link to Engadget
Basically there are a few new series that will be available this year
710 MLC-HET
720 PCIe-SLC
520 MLC
20GB Larsen Creek SLC?
Attachment 114040
Attachment 114041
Will read up on this later...
I was a bit surprised at this, it seems Intel ain't that serious on the SSD market -before I saw this^.
Q4-ish 11 yes, but atleast they're coming... XD
SLC is back! :eek:
And what the hell is MLC-HET? And why is it on SATA 3Gb/s while still being a part of 700 series?
questions is, will prices per gigabit be affordable?
Nice find :)
Looking at the ONFi 3 specs it seems that things will change a lot. AFAIK all the new gen drives are still at ONFi 2.
Maybe it's a reference to TLC?
http://www.pcper.com/news/Storage/Ne...hing-TLC-flash
I don't think so, TLC looks like a cost saving solution, not a performance one... But maybe you're right! Guess we'll see. :)
Maybe HET = High Endurance Tech? That would make sense... Enterprise solution. :)
HET = prefix hetero, heterostructure, heterogeneous and heterojunction, mainly hetero as the base process. No wonder will be SATA 3Gb/s.
TLC like Tables Ladders and Chairs ?
And maybe HET means Hyper Enhanced Threading. Well it's Intel you know... :p:
Really happy to see small capacity SLC-based ssd's in the roadmap...even if they are 3Gb/s
I guess the 520 series info will probably add fuel to the fire wrt Intel using the SF controller :)
(based on the capacities listed)
I expect HET is somehow related to Enterprise grade NAND.
Was kinda hopin' the 6G 520 would have a brand new BALLS-OUT intel controller laying the smackdown on everything. Oh well...
That SLC PCEe SSD is going to expensive.
hhmmm will be interesting i wonder if intel is going to have some of their own *new* controllers on these drives?
And the funniest thing is, ioXtreme is likely to still end up being faster... Or cheaper, at this rate. :p:
Should be exiting, though!
OT: is Fusion-io coming up with anything new? They've been pretty quiet for a while. Then again, I cannot access News section, so... :(
I don't see any reason to expect the 520 to use a Sandforce controller. It just looks like a 25nm flash update, plus additional capacity models, to the 510 which uses 34nm flash.
I think you are right about HET. It is probably another name for eMLC (enterprise MLC flash).
until they can get it to boot, i dont see any reason to. the stuff they have out now is the fastest out there, but without boot-ability. really is too bad too.Quote:
OT: is Fusion-io coming up with anything new?
@johnw
I noticed the typo, have fixed it.
I'm hoping Intel are making their own SSD controller but I can't see any signs of such a thing in the making, none of the other drives looks to have such a controller.
So, when those capacities are listed and they more or less fit perfectly with the typical capacities used on SF bases drives it will lead to speculations, don't read more into it :)
As for the fusion-IO stuff, the money is on Enterprise, not on "enthusiasts", at least for now.
http://www.intel.com/standards/nvmhci/
The interface has the following key attributes:
Does not require uncacheable / MMIO register reads in the command issue or completion path.
A maximum of one MMIO register write is necessary in the command issue path.
Support for up to 64K I/O queues, with each I/O queue supporting up to 64K commands.
Priority associated with each I/O queue with well defined arbitration mechanism.
All information to complete a 4KB read request is included in the 64B command itself, ensuring
efficient small random I/O operation.
Efficient and streamlined command set.
Support for MSI/MSI-X and interrupt aggregation.
Support for multiple namespaces.
Efficient support for I/O virtualization architectures like SR-IOV.
Robust error reporting and management capabilities.
The specification defines a streamlined set of registers whose functionality includes:
Indication of controller capabilities
Status for device failures (command status is processed via CQ directly)
Admin Queue configuration (I/O Queue configuration processed via Admin commands)
Doorbell registers for scalable number of Submission and Completion Queues
The goal is to help enable the broad adoption of SSDs using the PCIe interface. The NVM Express specification was developed by the NVMHCI Work Group, which includes more than 70 industry leading member companies. Core contributors include Cadence, Cisco, Dell, EMC, Fujitsu Technology Solutions, IDT, Intel, Marvell, Micron, Microsoft, Nvelo, Oracle, Pliant, PLX, Samsung, SanDisk, SandForce, STEC and Violin Memory.
''The NVM Express specification offers numerous benefits as a standard interface through the interoperability it fosters,'' according to the group. ''For example, each OS vendor may confidently write a driver that works for devices from multiple vendors. OEMs may procure devices from diverse suppliers that all implement a consistent feature set. Time to market may be reduced as validation times shrink with standard drivers and OEMs leverage the same test suites across devices.''
A standard Linux driver for NVM Express is already available. NVELO is developing a standard NVM Express driver for Microsoft Windows OS, with an alpha release scheduled for Q3 2011.
http://www.eetasia.com/ART_880063657...T_9fa6e66f.HTM
64GB is not a standard Sandforce capacity. It is a common Marvell capacity.
I did not suggest that the 520 would use an Intel controller. I was assuming that it would use a similar Marvell controller to the one in the 510. But that is just a guess, since the 520 seemed to be a replacement or update to the 510, judging from the chart. The other capacities, 480, 240, and 120GB seem to have more reserved flash than other Marvell controller SSDs. But then the 64GB capacity looks out of place among those others.
720 pcie slc well now that got me interested/my blood going
i dont care how much $ as long as its an intel controller
it better be intel controlled and not marvell
:sofa: Sorry guys, this was a wind up and I fell for it. Dang. - Deleted
RE: 720P... Does RAID5 provide any tangible benefit in this kind of situation? It's not like you can hot swap out some bad NAND or a failed controller and rebuild the array if something dies. Anyone running something like this is going to have a backup on mechanical anyway, so the parity seems to just be consuming pricey SLC NAND and controller CPU cycles that could be put to much better use IMHO.
Is the RAID5 there to improve the reliability of the SLC NAND? Is it error prone to the extent that this is necessary?
I believe if any 1 of the 4 fails, the array will just get rebuilt into a 3 device R0 array with similar performance as the original but with no more redundancy.
The 50 microsecond read latency of the 720P looks very promising. Slowly approaching FusionIO in this respect.
I honestly didn't notice that the 310 was SLC before. 20GB SLC drive for $110 seems like a pretty damn snazzy deal.
Larson (or Larsen) Creek is the 311 and according to the "roadmap" it is SLC.
The 310 has been out for some time and is MLC and is using the mSATA connector. (not compatible with the standard SATA connector)
Meh only 80GB as max in the next mSATA, give me 120GB for boot drive.
Quite a few of my setups use the Intel X25-V 40GB (or Kingston equivalent) as the boot drive, and I've still got 10-15GB available.
I'm using VMWare and so the boot drive doesn't need to be more than 32-40GB.
So, it's doable on small drives but it depends on usage.
I have had 60GB as primary (And only) for a long time and I am getting tired of it, but ofc if you use mSATA you will also have another drive, which will help.
But I just ordered a 128GB Max IOPS drive so wont need it anytime soon as I also have a normal drive in my W520 (Instead of dvd drive which I never use)
Yeah its the 311.
LinkQuote:
Originally Posted by VRZONE
I bet the 311 is SLC because Intel's new caching gizmo does a whole lot of writes to the SSD, since the most frequently accessed datasets would likely change a lot.
That may be a consideration, but I think the main reason is that even a 40GB Intel 320 SSD can only write at 45 MB/s, and the 311 is only 20GB so its write speed would be horrible if it did not use SLC flash.
I noticed from the picture of the 311 circuit board in Anand's review that the 311 does not appear to have the large capacitors that the 320 has for power-loss protection. Instead, there is a large "ISSI" IC that I cannot read the small print on. DRAM, I guess.
http://images.anandtech.com/reviews/..._DSC5607sm.jpg
Speaking of write speed, I wonder how well a 64GB Samsung 470 would do as a cache drive. It has the highest incompressible data sequential write speed of any MLC 64GB SSD, and the Intel chipset apparently limits the cache size to 64GB. But if 4KB random write, especially at high QD, is more important than sequential write speed, then the Samsung 470 would not be a great cache drive.
This should sort out the small print :)
Attachment 114293
I'm not sold on this new Z68 caching thing yet, might give the 311 a try though.