@killaW0lf04: really? so if you're right about the linux kernel - this would dramatically speed up the OS? Or not.. since this is more of a firmware program.

Maksimir

@Joseph: not necessaryly, page based file systems work very well with random access since there are direct pointers to file locations as far as i remember.

killaW0lf04

Samutz:

Linux will probably try include this in their next Kernel release. Windows will probably be quite hesitant since there quite intent on supporting NTFS for the long run. Mac OSX.........well im not so sure what apple would do in this case.

killaW0lf04

now work on improving the price 100x

winmac

I'm not sure I even understand when you'd have a lot of random writes. I'd think most of the time you want fast writes is in cameras, and pictures should all be sequential addresses. Doesn't really seem like there'd be any improvement there. The extra translation indexing will probably slow things down a trivial amount for sequential writes as well.

yertle

Yep writes go fast, reads go slow.

Joseph

This tech is so new, there will be so many speed related devs to come in the future in SSD. I'll buy when the transfer speeds hit the proverbial roof for once.

WaffleTeamStrike

@Samutz: The lack of coupling between physical and logical location makes it more like traditional drives.

Besides this appears to be something implemented on the SSD side rather than the OS side.

Log1c

Basically just like a RAID controller...

yertle

It should be like a black box to the PC. PC stores in logical addresses, which apparently have nothing to do with physical addresses. The SSD itself will translate between the two.

yertle

My question is: What OSes are going to support this new filesystem? I can't see Microsoft putting in the effort to make Windows compatible with it, especially with them being so dead-set on NTFS for the time being.

Samutz

@bobojuice: I would hope that you would NEVER defrag an SSD. They don't benefit from it. In fact, it reduces its effective lifetime.

zigziggityzoo

That little blurb doesn't explain the algorithm very well. It sounds like they just stuff the data wherever's most convenient. I wonder how this affects read times.

yertle

So basically, im going to have to defrag my system every 20 minutes?

bobojuice

@cardboredbox: Not likely. This is just reporting a file system improvement. That most likely has zero to do with hardware advancements. It will be interesting to see what operating systems will support it.

bucho54

Wait... that puts it up into the gigabyte range right?

Holy fuck that would be impressive, much more like solid state ram.

Log1c

So will this be a firmware update that can be sent to older SSDs, or are they going to save that for new models so people have to but another hard drive? If it's program based, it should be available to older hardware as well. I'm sure people who already have SSDs would really appreciate a 100x boost in write speeds.

GTgeek

So... will this make them cheaper??

cardboredbox

werd em up dawgy dawgy!

danman81

No wonder Samsung wanted to buy SanDisk.

daftrok

@bobojuice: I'm not quite sure where you got that assumption from, but it is the exact opposite. By using paging, theoretically this would eliminate any need for defragging.

The reason this hasn't been done before, is that it's easier for traditional hard drives to read sequentially (saves the drive from having to seek to another portion of the platter).

Since SSDs can index (aka page) things, given the appropriate file system that takes advantage of that concept, it is entirely possible to split the data up and access the parts wherever they reside on the chips. Hence the "no fixed coupling between physical and logical location" statement.

tandonmiir

@bobojuice: We defrag to increase sequential read performance, when the head doesn't have to move track to track, it spends more time pulling data off the disk. But for SSD, there is no head, so theres no reason to ever defrag it.

blitzcat

I'm not gonna pay a lot for this muffler er, SSD.

General Halfshaftery

@GTgeek: From reading about the tech, I would guess that a firmware update for older drives to give them this functionality wouldn't be possible. This will probably require a new set of hardware on the drive to pull off.

phor11

Extreme FFS? Haha. Best name ever.

Nickbee

will it allow more writes increasing the life of the disk?

raymondit

remind me to buy stock in Sandisk

spider2544

ok, fast random writes... but what about sequential read?... maybe on write it was simple because it writes where is faster, but on reads, you must read in the places where is the file...

but hey check this kind of dual core HDD ideea: (dual head HDD) [concepts.shpe.ro]

shpe11

@yertle: Probably not very much. Random access latencies on flash-based media are insanely low. As a result, even if the data is written sporadically all over the place (virtually, since it's really just different row/column pairs), the impact to both random and sequential read speeds and latencies will likely be minimal.

racermd

Probably some form on *nix. as somebody else said, these things usually get implemented into the linux kernal relatively fast

ack389

This is just a new FILESYSTEM, like fat 32, ext3, or ntfs. Soit isn't the 'firmware' that would be a problem, it is the OS. As of now probably no OS supports it. However most likely, some form of *nix will first

ack389

I wonder if it has any ties to the old Amiga's FFS (Fast File System)? Probably not...

KenK

i'm glad it's easy to get to the drive in my dell inspiron mini. by next year apparently i will be able to easily replace it with something much bigger and faster. very cool.

Aaron Stein

@KenK: If I were to hazard a guess, I'd also say probably not.

sxr7171

@yertle: Thank you. Now will all you people who think this has to be supported by the OS shut up.

sxr7171

@bobojuice:
if you're obsessive-compulsive that way, sure. not that you'd see any difference in performance.

willyolio

@cardboredbox:
Actually; yes it will(in a way).
With electronics, the more you produce, the less it costs to make(to an extent). Cost per a unit is directly tied to production volume.

So how does being faster make it cheaper? Because this gives an incentive for people to actually choose them over traditional HDD and thus will increase sales.
If they can make a drive with higher speed and reliability than the traditional "raptor" drive for a comparable price; they can replace that whole market and springboard their production volumes to a reasonable level that in turn could reduce the cost of production and open up new niches in the market as the price drops.

VishusBurn

@willyolio: Actually there is a difference in performance, it gets worse if you defrag it =P

TheLostVikings

I wonder how Extreme FFS compares with UBIFS, a file system specifically developed for raw flash that's already available for Linux.

Some SSDs have much faster random access speeds than others. For SSDs which have fast sequential read and slower random access, defrag might deliver some benefit.

skierpage

@yertle: There's no need at all for sequential addressing when you're using a solid state device. the sequential addressing thing only mattered when you were reading from a spinning disk, and it took a (comparitively) large amount of time to move the head.

ShamaHapiloons