Intel's Z68 Chipset to Bring SSD Caching to Sandy Bridge

[MattInOz]

This maybe a dumb question....

Why does the chipset support 4 DIMMS?
Don't all Sandy Bridge CPU have the memory controller on the CPU die, so isn't a memory controller in the Chipset a waste of space that could be used for other stuff?

 

[0815]

Kinda makes me glad I held out in getting an SSD for my MBP. It's still a very nascent technology and there's still tons of improvements being made over a short period of time.

I think I might just hold off on it for another 2-3 years when it's time to upgrade the computer. For sure SSDs will be likely mainstream by then.

Than enjoy your slow harddrive for the next 2-3 years. I upgraded the harddrive in my older MacBook Pro over a year ago to SSD - best upgrade every I made to any machine, everything starts instant, iPhoto is finally usable with big photo libraries. At work I have a newer and much faster (processor) MacBook Pro with old fashioned hard drive and it feels so damn slow compared to my old MBP with SSD. I only can recommend everyone not to wait a single day to upgrade to SSD, it's a huge difference. Prices will come down over time and drives will get bigger, but if you can effort it, do it - I don't need a faster processor, the harddrive is the bottleneck these days. (and I already put in a SSD upgrade request for my work MBP - sick of wasting time waiting for apps to start up)

 

[sparshdgr8]

It's not like they're going to incorporate this anytime soon anyway. This solution is handy because SSD's are still kinda expensive, but I think it would be better to bite the bullet and buy an SSD even if it's not that big.

can this new chipset be seen in these summers on the new MBP that will come with OS Lion?

 

[sparshdgr8]

can this new chipset be seen in summers on the new MBP ? Maybe, after the launch of OS Lion??

 

[adamtailor13]

caching SSD marries a conventional hard disk drive to a relatively small solid state (SSD), with the most frequently accessed automatically placed in the Quick Access SSD, while the two units appear to users simply as a single unit. The functionality serves to bring users the benefit of much of the speed of the SSD, but the storage capacity and lower cost of traditional hard drives.

 

[milo]

This is not a new concept, my Seagate Momentus does the same thing... It caches commonly used files to the 4GB SSD attached drive...

Similar but not exactly the same. Being able to use a separate hard drive and SSD and choose the size of each offers much more flexibility. The cache on the Seagate hybrid is way too small for the work I'm doing but a 128 or 256 SSD working with a TB or bigger drive would be great.

I'm not sure if it's a possibility, but this setup could potentially use one SSD drive to cache multiple hard drives.

 

[drsmithy]

Obviously nobody remembers Turbo Memory. This is just embedding it into the board.

http://en.wikipedia.org/wiki/Intel_Turbo_Memory

It did very little to improve performance and actually hurt performance in several areas.

http://www.anandtech.com/show/2252

I don't think it's really valid to draw conclusions on this from that. SSD performance - and in particular size - have come a long, long way since then.

This is basically an attempt to implement ZFS's L2ARC and (hopefully) ZIL in the chipset. IMHO this is the best usage of SSDs at current (and near future) technology levels and I've been waiting for other OSes/filesystems to implement it for years (Microsoft kind of did with Vista's/7's ReadyBoost, but you can't use regular SSDs and it doesn't cache writes).

Per those that say just get a SSD, this would still be beneficial. You could have small (4GB?) super fast SSD's (think almost RAM speeds) using the new caching tech, and have a regular-speed SSD (80GB) for your "regular" data. That would provide the best bang for the buck.

Better to just spend the money on 4GB more RAM.

This sort of thing really needs 10s or 100s of GB of SSD to be effective.

The real problem (from a marketing perspective) is that it's extremely difficult to benchmark, since it requires some time (anywhere from hours to days of representative usage) for the SSD cache to properly "warm up" before the performance benefits are seen.

Why do we need a new chipset to support SSD Caching? It seems like this could be done with software on the OS level... Windows can already do this, can't it?

The OS level is the best place to do it (because it knows about the filesystem, not just blocks), but OS vendors are dragging their feet. Vista and 7 have this somewhat implemented with ReadyBoost, but the real standouts are the OSes using ZFS.

One TB port may be fine for a laptop but but a couple of years down the road and desktops will find the need for more than one port. Laptops may never have that need but that is no surprise.

I disagree. Laptops are _exactly_ where TB can be most useful due to their physically constrained expansion options. I'm hoping future Apple laptops - particularly the Air - will soon come with at least two TB, if not three or more, TB ports.

(In the PC world it's somewhat less important since docking stations with things like multiple DVI outs, eSATA and even PCIe slots are available, but the same argument applies.)

Exactly the question I had. I assume a hardware solution may be optimal but even a software solution should be able to provide a big improvement over current solutions (similar to hardware raid versus software raid).

The "optimal" solution for both these things (caching and RAID) is in software, not hardware, because the OS has greater insight into the structure, type and frequency of data being accessed, and can hence make more intelligent decisions about how to improve performance.