Panasonic Working On 2-Terabyte SD Cards

An anonymous reader writes "SD cards with a theoretical maximum capacity of 2TB are in development by Panasonic and the SD Association, it has been announced. The technology is called 'Secure Digital Extended Capacity', or 'SDXC', and Panasonic has announced it will soon show off a 64GB SDXC card. Using the new technology, read/write speeds are set to hit 300MBps. SanDisk and Sony are using the same standard to develop Extended Capacity cards in Sony's Memory Stick Pro and Memory Stick Micro range. SDXC utilises Microsoft's new exFAT file system — AKA 'FAT 64' — which first appeared in Windows Vista SP1, and has a theoretical file size limit of 16 exbibytes." Reader xlotlu adds a note about the "proprietary exFAT file system, which is available for licensing under NDA. There are currently no specific patents on exFAT, but its legal status is uncertain since it's based on FAT. The FAT patents have been previously upheld in court."

Comments

They're talking about address space

[seanadams.com]

This article is absolute blithering bullshit. They're talking about the interface / file systems' _addressable_ size. Compared to actually achieving higher storage densities, that's about as hard as pulling a number out of the air. It has absolutely nothing to do with the technology needed to fit 2TB or any other number of bytes into whatever little card.

And oooh theyre making a 64GB card but "working on" a 2TB card? Yeah right, so only a 30-fold increase in density left to go!

Then he goes on to discuss throughput as if that has anything to do with it....

Re:They're talking about address space

[sexconker]

No! They're working on a 2 TiB card and have a 64 GiB card just about ready. They have a theoretical limit of 16 EiB!

Get with the times, accept the ibi!

(Fuck ibi. 1 KB = 1024 bytes.)
(Fuck this article. Might as well say Sony is working on a new battery that will recharge in less than 1 minute and last for days, while Intel is working on 32 nm CPUS, and later, 20 nm!)

Re:They're talking about address space

[Shrubbman]

No, the file system may support it but the SDXC standard will top out at 2TB. That way, after you buy all new kit this round they can get you to buy something else when they come up with their NEXT standard. It's called planned obsolescence, see the previous transition from SD to SDHC and this forthcoming transition from SDHC to SDXC. You really think they aren't planning to milk this cow every couple of years for as long as they can, rather than do it right and just come up with ONE standard that'll have headroom in the hardware logic to match that 16 EiB limit on the file system.

Re:They're talking about address space

[CaptainPatent]

No, the file system may support it but the SDXC standard will top out at 2TB. [...] It's called planned obsolescence

What are you talking about? 640k should be enough for anybody!

Re:They're talking about address space

[nine-times]

Yeah, I came in here to point out how silly this is. There's no announcement about any hardware in the pipeline. They're planning on using a filesystem built on FAT but with a 2TB theoretical limit? Who cares? There are better filesystems than FAT with theoretical limits much higher than 2TB.

In fact, the bigger question in my mind is, why is Microsoft coming out with a new version of FAT to support bigger filesystems? Wouldn't the effort be better spent on figuring out how to kill FAT once and for all and replace it with something that doesn't completely suck?

Re:They're talking about address space

[CaptainPatent]

No! They're working on a 2 TiB card and have a 64 GiB card just about ready. They have a theoretical limit of 16 EiB!

very doubtful actually. With a magnetic media, the odd division of the disk into sectors nearly always meant each new layout was a different, non-standard size. Drive manufactures in that case found every way possible to round up, including using 1000 bytes as a KB when it is only a KiB. Flash forward to... well... flash. When making the circuitry, it actually takes less work and programming (generally) to round out all address spaces to a given digit to be used. In addition there are very standardized chip memory sizes for flash. I don't know specifically if Panasonic is going to disregard the standard chip sizes and only shoot for KiB, but that's the case far less frequently with flash than with standard rotating hard drives.

Re:They're talking about address space

[andy_t_roo]

640k may be enough, but only if not many people talk about it.
I think that some sets of comments here on /. could end up being > 640k -- the 36 comments that this page is up to take 112k.

Re:They're talking about address space

[cheater512]

It could be *much* worse.
It could be based on NTFS.

I still cant understand why Microsoft is the only company who hasnt been able to make fragmentation resistant file systems
Hell OS/2 had HPFS which didnt fragment and that was ages ago and made partly by Microsoft.

Or maybe they desire the 'gradual slow down' effect that fragmentation causes.
So they release a new version of Windows just in time and advertise that its even faster, and it does actually seem faster.

Re:They're talking about address space

[sexconker]

7680x4320 (8K) x 120 (fps) x 2 (3D) x 3 (lame-o 24 bit color) = 22.247314453125 GB/sec. 1 hour is 78.213214874267578125 TB.

Oh, you meant compressed? I don't accept compression at the source level!

Re:They're talking about address space

[sexconker]

...
The 1000/1024 fiasco is due to marketing, nothing more.

Re:They're talking about address space

[sidyan]

Drive manufactures in that case found every way possible to round up, including using 1000 bytes as a KB when it is only a KiB.

Actually, the size of a KiB (Kibibyte) is not in question, it's most certainly 1024 bytes. The kB (Kilobyte) on the other hand, is used by drive manufacturers to mean 1000 bytes, and in this they are only following IEC, IEEE and ISO standards.

Re:They're talking about address space

[jmorris42]

> Wouldn't the effort be better spent on figuring out how to kill FAT once and for all and
> replace it with something that doesn't completely suck?

1. NTFS is to complex and undocumented to be used in embedded consumer electronics.

2. Microsoft needs to keep control over the file system used in consumer electronics. If they hadn't offered this up (for a small fee of course) vendors might have been forced to look elsewhere... at the many filesystems in Linux or BSD that easily scale to the sizes required and have free reference implementations available, although the GPL would preclude many embedded vendors from directly using many of the more popular ones's code.

If I had to guess Microsoft will give em a sweet deal on the license fees so long as they give desktop linux some patent hell on implementing support, thus allowing SuSE to ride their trojan horse again.

And from the 2TB upper limit I'm guessing the are not reworking the maximum block size so there will still eventually have to a "LBA48" style incompatibility breakage at some point. Because 2T on a full size SD card isn't decades away.

Re:They're talking about address space

[geniusj]

Don't worry.. In the coming years, fragmentation won't matter nearly as much. On will come the log-structured filesystems and their ilk to replace the heavily disk-tuned mainstream filesystems we use today.

Re:They're talking about address space

[grumbel]

Correct me if I am wrong, but didn't the standard came *after* the HDD manufactures used to abuse the 1000/1024 for their benefit?

Re:They're talking about address space

[sentientbeing]

Peter Piper, from the Peck of Pickled Pepper organization.

Re:They're talking about address space

[fuzzyfuzzyfungus]

Most scenes aren't actually as entropic as all that, so lossless compression could almost certainly be applied. In particular, I'd be absolutely shocked if, in any but contrived scenarios, a stereoscopic image pair requires anything close to double the storage of a single image.

Your point is valid in that 2TB isn't actually all that much, for really tweaky video; but lossless compression should be considered.

Re:They're talking about address space

[vlad30]

Normally I'd agree. But after SDXC, they've run out of good options for acronyms.

SDUC (Ultra) SDxC (xTreme!) SDSC (Super)

SDPL (Plaid) They'll go Plaid

Re:They're talking about address space

[Anonymous Coward]

I still cant understand why Microsoft is the only company who hasnt been able to make fragmentation resistant file systems
Hell OS/2 had HPFS which didnt fragment and that was ages ago and made partly by Microsoft.

Bullshit. Many filesystems can and do experience fragmentation; fragmentation in NTFS is only such a big deal because of the existence of defragmentation tools. One of the new features of ext4 is an online defragmenter; there was an ext2 offline defragmenter, but nothing for ext3. XFS has an online defragmenter as well. OS X's HFS+ also does limited automatic defragmentation.

Re:They're talking about address space

[YrWrstNtmr]

You insensitive clod. I can already fit all my porn on one SD card.

Slacker.

Re:They're talking about address space

[biryokumaru]

Acutally, that's not necessarily true. Exabyte is sometimes 2^60, sometimes 10^18. That's why the "ibi" system was developed: to be specific. The old system allows marketing folks to cheat people. At the "exa" size, the difference is more than 15%.

Re:They're talking about address space

[karnal]

All of a sudden, Emeril Lagasse appeared from the ether and punched your hard drive with "extra fragmentation." *BAM!*

Re:They're talking about address space

[Lorien_the_first_one]

You're right about that, except for the patent hell part. The proprietary part will reverse engineered in short order and MS will be afraid to sue. Why? Eben Moglen points it out very nicely here in the "Be Very Afraid Tour": http://www.youtube.com/watch?v=6YExl9ojclo So while it's true that MS and Novell have a deal that could protect their customers, the success of MS' patent threats depend on the ignorance of the companies that could be threatened by them. And besides, once a lawsuit is filed, MS is going to have a lot of public explaining to do in discovery. And their patents will still have to withstand recent rulings. Its only a matter of time before it will become politically and economically untenable for MS to use those patents against Linux.

Re:They're talking about address space

[Joce640k]

They we all know WHY it was invented, we just want the name/address of the person who thought it rolled elegantly off the tongue.

Re:They're talking about address space

[Anonymous Coward]

SDEC (Expanded)

Or they could go the Capcom way with Super Secure Digital II Turbo Hyper Cheating Fighting Champion Edition Capacity.

Re:They're talking about address space

[Linker3000]

I use micro-SD cards for my midget porn!

Re:They're talking about address space

[gbjbaanb]

You insensitive clod. I can already fit all my porn on one SD card.

Slacker.

The word you're looking for certainly ends in 'ker', but it isn't the one you used :-)

Re:They're talking about address space

[rufus t firefly]

Boastful, forgetful, a fucking liar? Who knows.

When you've got his kind of money, you can buy whatever kind of truth you want, apparently.

2TB? exFAT?

[Anthony_Cargile]

Well even if I would take so many pictures on my camera that I'd need twice the size of the library of congress to hold them all, not too happy about some proprietary filesystem (assuming it isn't ro/rw on all platforms yet).

But still, I would buy one just so I could take it out of my pocket whenever I was having a problem so I could say, "Well, this was possible, so...." despite never using it.

</humor></criticism>

Re:2TB? exFAT?

[Seth Kriticos]

The problem with ext3 is the same as with NTFS. They are journaling file systems, meaning that they do a lot of read/writing to the media. This is no problem with hard drives but flash media have just a limited amount of these. That's why non journaling file systems are employed on them. Mainly some kind of VFAT but ext2 would also do the job quite nicely. Problem is, that ext2 is not supported on the Windows platform as a default option. Additional installation is not an option for the avarage Joe. The bigger media will have some sort for FAT for the same reason. I'm qurious, is this new version of FAT also readable on non Windows platforms? Anyway, if you feel like it, it should be no problem to reformat them to ext2. Just your Windows friends will look a bit surprised that it seems unformatted. If you go more into the thing, you probably can also partition that thing like normal USB flash media, put a standard VFAT in one partition to it's limit and fill the rest with ext 2 to 'hide' your info from your clueless friends / boss / whatever. Or you put an array of normal FAT partitions on it with their maximum limitations. Did I miss something?

Re:2TB? exFAT?

[ion.simon.c]

Ext2.

There are plenty of available drivers.
If MSFT wanted to roll their own, they could base their ext2 driver off of the one of the ones in one of the BSDs. Why the fuck is MSFT reinventing the wheel again? :(

Re:2TB? exFAT?

[denzacar]

Think Personal Life Recorders.

Stick 2TB card in a audio/video/photo/data/navigation/internet device - record EVERYTHING.
You could store about a years worth of video in more than a decent quality PLUS have a shitload of space left for everything else.

Attach a portable viewing device and make nearly every form of artistic entertainment delivery model obsolete.
Movies? In theaters? Why? There are going to be so many cams on the day 0 that SOME of them must be watchable.
Most people have no sense for quality anyway. And at ~1GB per movie you could have about 160 DAYS of non-stop video content on 1 card.
Get a couple of cards and you could have every movie you will ever see in the palm of your hand.
About a 100 cards will record your entire life anyway.

Music? Sure. Paying for it? Umm... again.. why? For a CD? You like the artist - send him a buck or two through his paypal account.
Every single live event in the world will soon be captured by someone, somewhere.

Books? Scannable by just flipping the pages.

Lectures? Seen one, seen em all.

Privacy? Very soon non-existent.

What we will start seeing a shitload of as a result?
Various digital watermark/thumbprint technology in EVERYTHING. And simple ways to bypass them.
No jewelery zone suggestions and attempts in enforcing.
Criminalization of EVERYONE, EVERYWHERE and resulting paranoia.
Accountability of everyone, everywhere for everything they ever do. Personally, I would love to see something like this mandatory for politicos.
Clash of need to spy on everyone, everywhere by governments with entertainment lobbies that would like to ban any and all digital recording.
A shitload of amateur porn.
A lot more problems with (better) identity theft.
Death of keys, passwords, and any locking system not featuring encrypted data on a chip not readable by a human.

End of the world as we know it.

Re:2TB? exFAT?

[Hal_Porter]

exFAT has been reverse engineered by some Japanese guy here

http://bbs.znpc.net/viewthread.php?tid=5366&page=1#pid33197

The downside is that he seems to have used Google translate to translate his document into English. I hexdumped some exFAT volumes and he seems to be correct about what the fields in the boot sector and directory entries mean.

Executive summary. exFAT has, at least when Vista formats a disk, one FAT that is basically the same as FAT32 except that all 32 bits of the entries are used rather than the upper 4 bits being reserved.

The boot sector layout is new and the main difference that the volume size in sectors is now a 64 bit entry. Cluster size is a byte allowing upto 2^256 sectors per cluster, though the Microsoft implementations have a limit of 32MB.

So you could have 2^32 clusters, upto 32MB each. For a reasonable cluster size of 64KB you can have 2^48 bytes or so sized filesystems.

Directory entries have a new layout with no short filenames. The file size is 64 bit, to allow files bigger than 4GB, which is the most pressing limitation with FAT32. There are also special nameless files. One is a bitmap of free clusters, the other is a table to convert Unicode characters to uppercase, since this is part of the filename hashing.

Adding a bitmap is to speed up free space. Finding free space in FAT requires reading the FAT until you find a free cluster. Each fat entry is 32 bits, and the FAT would grow to 16GB on a 2^32 cluster volume. Adding a bitmap with one bit per cluster makes the worst case read a mere 512MB, perhaps 10 seconds or so on a modern drive. Of course the normal case is that you search from the last cluster allocated and find a free cluster much quicker than this. In fact you could cache the bitmap and most likely you find a sector in the same cache line that you found the last one. With a smaller device of course, the bitmap will be smaller too. If you were really feeling adventurous you could compress the bitmap into extents of free clusters and use those to allocate free space instantly most of the time.

So it's more or less FAT32 with the legacy stuff removed, volume size in sectors and file size in bytes fields widened from 32 to 64 bits and a cluster allocation bitmap ( Tanaka-san calls this KURASUTABITTOMAPPU, a sort of Japanesified version of cluster bitmap ) added to speed up free space searches. You can have volumes of upto 2^32 clusters or 2^64 sectors and files of upto 2^64 bytes.

Windows CE supports a transactional version of exFAT. I'd guess this has two FATs and switches between them to get transaction safety.

Incidentally, if this ends up part of something like SDXC it will be quite cheap to license, so I'd expect consumer electronics devices that need to support big files to start to use it as well as or even instead of FAT32. It wouldn't be hard to write a driver that supports both. The Microsoft exFAT driver in Vista works fine on XP. I'd guess someone will write a Linux driver once the reverse engineered internals become well known. It's not clear what patents Microsoft have on this. My guess is that they won't sue end users for using it, but they may approach Linux vendors like they did with Novell.

FAT

[serviscope_minor]

The only, ONLY good thing about FAT is that it is very well understood and supported everywhere. Why on earth would someone license a proprietary filesystem based on an awful filesystem when they don't need to?

Oh, and why on earth would a SD card manufacturer need to license a filesystem in the first place? It's not like it'll care what data is on there.

Re:FAT

[Amouth]

because they ship them formatted because 90% of buyers are stupid.

it is cheaper for them to pay a fee to licence a file system - preformat the card than it is to ship them blank only to have the consumer call in saying it is broke + the costs associated with said consumer returning perfectly good products thinking they are broken.

users are idiots - if they ship them non formatted then people will think they are broken

- yes it's sad - yes its true

Re:FAT

[More_Cowbell]

- yes it's sad - yes its true

Well, it certainly is sad, and most likely true to some extent, but seriously I'm only 31 and I recall easily the days when all disks were shipped unformatted. I would like to think that ^most^ consumers could get the hang of formatting disks fairly quickly.

But then with ^most^ users using Windows, wouldn't they format it with FAT anyway?

Re:FAT

[DoofusOfDeath]

because they ship them formatted because 90% of buyers are stupid.

Stupid != not wanting to waste mental bandwidth on how to prepare a digital medium for use.

The world is too amazing. and life is too short, for all intelligent person to waste time worrying about that crap.

Re:FAT

[TheRealMindChild]

Don't let the name fool you. This is less a "FAT" file system than FAT32 was to FAT16. It barely resembled anything FAT AT ALL. Long file names are different. Storage is different. The boot sector layout is different. File entries are different.

A snippet from wikipedia (since I can't find a link to the specification right now):

exFAT is an incompatible replacement for FAT file systems that was introduced with Windows Embedded CE 6.0. It is intended to be used on flash drives, where FAT is used today. Windows XP file system drivers will be offered by Microsoft shortly after the release of Windows CE 6.0, while Windows Vista Service Pack 1 added exFAT support to Windows Vista. exFAT introduces a free space bitmap allowing faster space allocation and faster deletes, support for files up to 2^64 bytes, larger cluster sizes (up to 32 MB in the first implementation), an extensible directory structure and name hashes for filenames for faster comparisons. It does not have short 8.3 filenames anymore. It does not appear to have security access control lists or file system journaling like NTFS, though device manufacturers can choose to implement simplified support for transactions (backup file allocation table used for the write operations, primary FAT for storing last known good allocation table).

Re:FAT

[CannonballHead]

You're right. When I buy something, I don't want to be able to plug it in and use it. I want to have to sit for $X $unit_of_time before being able to put my files on it!

What's this with computers coming pre-built and even pre-installed with $operating_system_name, anyways? I don't want to plug it in and have it work, I want to have to spend $X $unit_of_time before I'm able to use it!

... your point remains true that people would think it's broken, but I'm not sure that "smart people" wnat to have to format it, either.

Re:FAT

[MrCrassic]

Stupid?

1. Pop the card in...
2. Windows tells you that it's unformatted...
3. You format it...
4. You're done!

Re:FAT

[Anonymous Coward]

You're talking to an audience of people who would buy shirts with RTFM in bold caps on them.

While a part of it is a smug hubris, we do wear them to make a statement about people who "don't have the time" to read simple instructions or a guide. We especially hate it when people with that ideology then start to run their fucking mouths off about how much something sucks because they had to read a manual. We also hate it when by their din of whining cries we are forced into industry standards that treat us like lazy morons.

For that lazy consumer who is the reason we cannot have nice things I wear that shirt. Also, NO I WILL NOT FIX YOUR COMPUTER!!!

Re:FAT

[slimjim8094]

No. Stupid is returning a product that works because you're too stupid* to read the back of the package that says THIS MUST BE FORMATTED.

All digital cameras have a format option.

*And I meant stupid, not ignorant. Ignorant would be 'well I didn't know I had to do that, but now I do'. Stupid is not even checking.

Re:FAT

[innocence18]

I don't want to put it in my machine first. I want to put it in my camera and start taking photos. Unless every single possible device I could put it in lets me format it, then shipping it unformatted is completely unacceptable.

Re:FAT

[PitaBred]

Back when disks were shipped unformatted most people didn't have computers, only geeks and technologically oriented people did. Which meant they could read a manual. When your market expands, so does the bell curve, and with it come the people at further ends of the capability spectrum.

Re:FAT

[FireFury03]

Surely the point is that with various different file systems the chances are quite high that you'll insert SD-shaped-card-with-new-disk-format into your consumer device, it'll try and read it and assume it's unformatted, and pressing Y at your prompt results in it trying to write FAT32 all over it, wiping the contents?

Assuming the camera pops up that message whenever it can't read a card, that's going to happen whether cards come preformatted or not.

On the other hand, it isn't exactly hard to tell whether a card is unformatted (all the bits are on or all the bits are off) or has data on it that you don't understand - looking at the first 512 or 4096 bytes should give you a pretty good clue. So the camera should really only present a "this card is unformatted" message for blank cards - if the card has data on it then something more appropriate should be displayed, such as "this card can't be read by the camera, it may contain data in another format. Do you want to format it and erase this data?". Yes, some people will still press yes and lose all their data, but there is a limited amount you can do to protect idiots from themselves.

Re:FAT

[Detritus]

Some disks have always been shipped formatted because they were used in systems that could not format their own disks. That was common for 8" floppy disks. See the DEC RX01 for an example.

They tried RieserFS

[Anonymous Coward]

But it died on the table.

I, for one...

[banffbug]

cannot wait until everyone's walking around saying "pebibytes". Thou some people will call them pebeebytes and that will annoy me to no end.

Re:I, for one...

[banffbug]

Please forget what i just said. I am wrong so so wrong.

Waste of time

[ickleberry]

Since everyone is getting rid of DRM anyway would it not be a good idea to drop the SD standard altogether and continue where MMC left off? a bit like the way Sony are getting rid of MagicGate

SD is just a RIAA-approved version of MMC with extra DRM features added. Maybe I'm just a bitter old sod but I find this continuation of the SD standard and it's DRM suspicious, perhaps they are waiting for a good time to re-introduce DRM on a massive scale and since every SD card ever made already supports it they will have no problem implementing it

I bet most the supposedly hardcore RIAA-hater nutjobs don't even realise SD has the built in DRM. They have been selling DRM-enabled cards for about 10 years now and just because the SD DRM hasn't seen any widespread use nobody batts an eyelid.

Re:Waste of time

[marcansoft]

SD is more than MMC+DRM. It added the 4-bit protocol which is pretty different from the SPI-style that MMC used and which helped improve transfer speeds. There are also quite a number of changes to the protocol. The DRM seems to be pretty worthless anyway - does anyone actually use it?

On a sidenote, SDHC already has a maximum addressable space of 2TB (2**32 512-byte sectors), though it's limited to 32GB purely artificially by the wording of the spec. Methinks this is mostly marketing and not a real change.

More proof it's too late for copyright.

[w0mprat]

With the continual increase in the capacity of storage, exponentially decreasing cost per size, and ever increasing bandwidth to link it all together, I wonder if there is there any use worrying about piracy.

You could say piracy moved to the internet because floppy disks were useless and CD/DVD burning costly, even when it's now rather cheap. Generally piracy has been scaling with availability of bandwidth and storage. But is there a point where it gets so stupidly cheap and powerful that old world business models become completely untenable?

Re:More proof it's too late for copyright.

[moogsynth]

I've been pilfering the mp3 collections of everyone I know via meatspace for some time now, and have a pretty massive collection stored over several drives. I call it my INFINITY JUKEBOX! People come to me when they want music, rather than get it online. I've given complete mirror images of the drives to a couple of people already. When it gets to the stage where all this data can fit on to a single shitty card, hell, I'll hand them out in the street for free.

I'm pretty sure I'm not alone in this.

ZFS?

[argent]

Given that ZFS has been optimized for flash, why bother with FAT?

Crazy to use exFAT

[macemoneta]

Since Linux, Mac and even most existing Windows users won't be able to use exFAT/FAT64 formatted media, they're not doing anyone any favors.

They could use NTFS as a more common file system, except for that whole journaling burning up the flash thing.

The most reasonable alternative is ext2, though I wouldn't want to spend a day fscking a 2TB SD card any more than I'd want to spend a day with chkdsk on an exFAT formatted one.

If flash sizes are going to continue to grow, they need to deal with journaling filesystems. Perhaps the easiest, most cost effective way to do this is by pre-partitioning the unit, with the bulk of the storage in one partition, but a second partition for a much smaller external journal aligned to more robust flash (e.g., 128MB with a 50M+ write life). Even with a 5 second journal update interval, that would give you about 8 years of 24 x 7 x 365 usage. Ext3 supports this configuration, not sure about NTFS or HFS+.

Microsoft does not own ZFS

[SuperKendall]

No license stream if you pick a sensible filesystem, sorry. Instead you get Microsoft further extension of FAT. Ack.

SDHC can already do 2TB

[Anonymous Coward]

Says Wikipedia "the SD 2.0 standard in SDHC uses a different memory addressing method (sector addressing vs byte addressing), thus theoretically reaching a maximum capacity of up to 2 TB (2048 GB). However the SD Card association has artificially defined the maximum limit of SDHC capacity to 32 GB"

Sounds like another way to extort people into using MS only standards. Hooray!

What the hell are you talking about?

[Sycraft-fu]

Unless you have a file system that simply refuses to do a write when there is no contiguous block large enough you WILL have fragmentation. Go do some brief searches if you don't believe it. For example one of the features ext4 will have over ext3 is an online defragmenter, meaning you can defragment the volume while it is in use (as Windows defragmenters do).

Re:Give me a break

[Ksevio]

I didn't realize how slow DSL really was if it's taken you over 2 months and you haven't even filled up half a gig. Glad I got cable instead

FAT 64

[fru1tcake]

a theoretical file size limit of 16 exbibytes

That's almost enough to store a picture of yo mama!

Re:FAT64 only for digital cameras

[maxwell demon]

Of course, if you don't use it on a device like a digital camera, there's no reason to use an SD card at all.