Searching \ for '[OT] 64GB disk on key' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=64gb+disk+key
Search entire site for: '64GB disk on key'.

Exact match. Not showing close matches.
PICList Thread
'[OT] 64GB disk on key'
2007\05\29@210337 by Peter P.

picon face
I did not know that they got so big (and so expensive). I think that the end of
the hard disk era is coming nearer with this. This disk on key is only 5 times
smaller than a 'standard' hdd supplied with new systems now. If Moore's law etc
holds in three years there should be TB size flash disks on the market. With a
lot of laptops and portables switching to a disk-less system is a strong
incentive imho.

 http://www.buslinkbuy.com/products.asp?sku=BDP2%2D64G%2DU2

Peter P.


2007\05\29@214346 by Jake Anderson

flavicon
face
Peter P. wrote:
> I did not know that they got so big (and so expensive). I think that the end of
> the hard disk era is coming nearer with this. This disk on key is only 5 times
> smaller than a 'standard' hdd supplied with new systems now. If Moore's law etc
> holds in three years there should be TB size flash disks on the market. With a
> lot of laptops and portables switching to a disk-less system is a strong
> incentive imho.
>
>   http://www.buslinkbuy.com/products.asp?sku=BDP2%2D64G%2DU2
>
> Peter P.
>
>
>  
Those guys are really expensive too as far as I have seen.
I can get 16Gb for ~$200 at the local computer market, I am half tempted
to make up a RAID system for the missus to play her games on out of a
few of them.

2007\05\29@230114 by Gökhan SEVER

picon face
They are continuously expanding on size, but what about the USB 2.0's speed?

Are they capable enough to replace a typical laptop's hdd?

On 5/29/07, Jake Anderson <spam_OUTjakeTakeThisOuTspamvapourforge.com> wrote:
{Quote hidden}

> -

2007\05\29@232539 by Marcel Birthelmer

picon face
On 5/29/07, Gökhan SEVER <.....gstr2005KILLspamspam@spam@gmail.com> wrote:
> They are continuously expanding on size, but what about the USB 2.0's speed?
>
> Are they capable enough to replace a typical laptop's hdd?

I think USB speed will be fast enough. The main problems will be flash
(probably NAND?) speed and rewriteability. If an OS decides to heavily
use the disk for temporary files etc., those few hundred thousand
write cycles can be used up very quickly.
- Marcel

2007\05\29@232720 by Harold Hallikainen

face
flavicon
face
How many times can you write to a specific byte? How fast can you write?

Harold


{Quote hidden}

>> --

2007\05\30@024632 by Tamas Rudnai

face picon face
...also, did you see the price? For that I could buy rack of ultrawide scsi
storage which spinning at 15krpm. Anyway, my first HDD was 40MB and now in
my phone there is a 2GB M2 card which is smaller than the nail on my little
finger.

Tamas


On 5/30/07, Harold Hallikainen <.....haroldKILLspamspam.....hallikainen.org> wrote:
{Quote hidden}

2007\05\30@030459 by Ariel Rocholl

picon face
Correct me if I am wrong, but I do not think Moore's law applies to memory
technology, it is more related to uP and its capabilities, remarkably speed.
Memory capacity nor speed have followed Moore's law AFAIK.

2007/5/30, Peter P. plpeter2006spamspam_OUTyahoo.com:
>
>  If Moore's law etc holds in three years there should be TB size flash
> disks on the market.


Peter P.
--
Ariel Rocholl
Madrid, Spain

2007\05\30@035400 by Peter P.

picon face
Harold Hallikainen <harold <at> hallikainen.org> writes:

> How many times can you write to a specific byte? How fast can you write?

The device has a datasheet on the webpage. They say write is 15 MB/sec and reead
20 MB/sec. This is 'quad USB 2.0' technology (they say so). If it's true (they
could be Mega Bits) it's faster than a current laptop drive by a factor of 2 or
3.

Also combining several drives using an 8-er USB hub to make a RAID is really
easy and hot pluggable. Under *nux with RAID on LVM slices adding units simply
adds capacity without requiring any system configuration manipulations.

And with that speed (and assuming a good USB infrastructure) a striped RAID can
probably exceed 100 MB/sec write speed without even trying hard. Even at 15
MB/sec an image copy at 64GB takes about an hour. That is faster than the
required time for a hdd of the same size.

Peter P.


2007\05\30@035529 by Peter P.

picon face
Ariel Rocholl <arocholl <at> gmail.com> writes:

> Correct me if I am wrong, but I do not think Moore's law applies to memory
> technology, it is more related to uP and its capabilities, remarkably speed.
> Memory capacity nor speed have followed Moore's law AFAIK.

I don't know if Moore's law applies here but installed disk capacity and RAM
capacity in commodity consumer computers roughly doubles or triples every 2
years without increasing the price significantly. CPU speed does not follow this
'law' mostly because there are physical limits that have been reached (instead,
double and quad cored units appeared). Of course I was being optimistic but
size-wise there is no reason to reach TB sizes in a few (2-3-4) years.

Peter P.


2007\05\30@041513 by Peter P.

picon face
Peter P. <plpeter2006 <at> yahoo.com> writes:

> double and quad cored units appeared). Of course I was being optimistic but
> size-wise there is no reason to reach TB sizes in a few (2-3-4) years.

That would be 'no reason not to reach' of course.

Peter P.



2007\05\30@095320 by Tony Smith

picon face
USB 2.0 is faster than most LANs (480 vs 100Mbps).

Tony


{Quote hidden}

2007\05\30@102110 by Michael Rigby-Jones

picon face


>-----Original Message-----
>From: KILLspampiclist-bouncesKILLspamspammit.edu [RemoveMEpiclist-bouncesTakeThisOuTspammit.edu]
>On Behalf Of Tony Smith
>Sent: 30 May 2007 14:53
>To: 'Microcontroller discussion list - Public.'
>Subject: RE: [OT] 64GB disk on key
>
>
>USB 2.0 is faster than most LANs (480 vs 100Mbps).
>
>Tony

But significantly slower than Ultra ATA, and much, much slower than the next generation SATA standard (3Gbits/s).

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2007\05\30@102646 by Matthew Mucker

flavicon
face
USB is, yes, but are the flash memories on the devices?

My experience is that (locally attached) mechancial hard drives are
SIGNIFICANTLY faster than USB key drives. For this reason I don't see flash
memories replacing hard drives any time soon.

{Original Message removed}

2007\05\30@103209 by Gerhard Fiedler

picon face
Marcel Birthelmer wrote:

>> Are they capable enough to replace a typical laptop's hdd?
>
> I think USB speed will be fast enough. The main problems will be flash
> (probably NAND?) speed and rewriteability. If an OS decides to heavily
> use the disk for temporary files etc., those few hundred thousand write
> cycles can be used up very quickly.

If that's the only problem, then just put the temporary files on a RAM
disk. Most current OSes have already versions or extensions that deal with
this and buffer writes in RAM.

Anyway, the moment Flash becomes a viable choice for a typical PC, the
already existing OS extensions that handle Flash become mainstream, too.
There's no reason why this couldn't (and wouldn't) be done. It's just a
matter of Flash price vs. HDD price. And probably not far away.

Gerhard

2007\05\30@105054 by Tony Smith

picon face
> >USB 2.0 is faster than most LANs (480 vs 100Mbps).
> >
> >Tony
>
> But significantly slower than Ultra ATA, and much, much
> slower than the next generation SATA standard (3Gbits/s).


Of course, internal is always faster.  Internal flash won't be as fast as a
hard drive (yet), but for a laptop power usage is a big consideration.

Flash drives will be rather quiet though (media PC).

I work for a company that redirects %temp% to a network drive.  That slows
things down a bit too.

Tony

2007\05\30@122526 by Sean Schouten

face picon face
On 5/30/07, Tony Smith <spamBeGoneajsmithspamBeGonespamrivernet.com.au> wrote:
>
> > >USB 2.0 is faster than most LANs (480 vs 100Mbps).
> > >
> > >Tony
> >
> > But significantly slower than Ultra ATA, and much, much
> > slower than the next generation SATA standard (3Gbits/s).
>
>
> Of course, internal is always faster.  Internal flash won't be as fast as
> a
> hard drive (yet), but for a laptop power usage is a big consideration.
>
> Flash drives will be rather quiet though (media PC).



Tony is right. Depending on the application, speed is not always an issue.
Regular A/V streams (think media PC) be just fine off of a flash-disk. It's
like me not needing a Porsche to do the shopping in...  ( Well, actually I
do. I just don't have the money for one -yet-! As the same money-issue also
effects me buying a beefy flash disk for my PC, I'll stick with the faster
and CHEAPER HDU's for a while! Poor me! )

Has anyone heard of FeRAM? It sounds promising; it's probably expensive
too...     http://en.wikipedia.org/wiki/Ferroelectric_RAM



Sean

2007\05\31@045844 by Peter Bindels

picon face
On 30/05/07, Michael Rigby-Jones <TakeThisOuTMichael.Rigby-JonesEraseMEspamspam_OUTbookham.com> wrote:
>
>
> >-----Original Message-----
> >From: RemoveMEpiclist-bouncesspamTakeThisOuTmit.edu [piclist-bouncesEraseMEspam.....mit.edu]
> >On Behalf Of Tony Smith
> >Sent: 30 May 2007 14:53
> >To: 'Microcontroller discussion list - Public.'
> >Subject: RE: [OT] 64GB disk on key
> >
> >
> >USB 2.0 is faster than most LANs (480 vs 100Mbps).
> >
> >Tony
>
> But significantly slower than Ultra ATA, and much, much slower than the next generation SATA standard (3Gbits/s).

Allow me to sell you a floppy drive that sends its data over a 10GB/s
ethernet connection. That'll be fast!

2007\05\31@122037 by Michael Rigby-Jones

picon face


{Quote hidden}

It might be if is has a big, high speed cache ;)

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================


'[OT] 64GB disk on key'
2007\06\02@073012 by Howard Winter
face
flavicon
picon face
Tony,

On Wed, 30 May 2007 23:52:42 +1000, Tony Smith wrote:

> USB 2.0 is faster than most LANs (480 vs 100Mbps).

That's the raw data rate, not the actual throughput.  I have done tests with USB2 (480kbps) and Firewire (400kbps), and the nominally faster USB2
takes longer to copy a given file than Firewire (all else being equal).

And of course there's Gigabit ethernet - I have a Lacie "Ethernet Mini Disk" which has a gigabit interface and a USB2 interface too - I'll try the
experiment both ways when I get some time.

Cheers,



Howard Winter
St.Albans, England


2007\06\02@124639 by Tony Smith

picon face
> > USB 2.0 is faster than most LANs (480 vs 100Mbps).
>
> That's the raw data rate, not the actual throughput.  I have
> done tests with USB2 (480kbps) and Firewire (400kbps), and
> the nominally faster USB2 takes longer to copy a given file
> than Firewire (all else being equal).
>
> And of course there's Gigabit ethernet - I have a Lacie
> "Ethernet Mini Disk" which has a gigabit interface and a USB2
> interface too - I'll try the experiment both ways when I get
> some time.


Sure, there's always something faster, but that wasn't the point.  

In a lot of cases you could replace the hard drive in laptops with flash
(USB or not) and I doubt most people would even notice.  They might notice
less noise, more battery life and sticker shock, but speed isn't high on the
list these days.

I've worked for companies where everyone used laptops & WiFi, a USB 'hard
drive' will be faster than that.

Tony

2007\06\02@125748 by Sean Breheny

face picon face
Has anyone addressed the problem of max number of writes, though? Most
modern OSes use virtual memory extensively. Having to cache the flash
drive to prevent memory cell wear-out due to writes defeats the
purpose of virtual memory, since you would need almost as much cache
memory (I think) as the virtual memory size.

Also, I thought that flash achieved its relatively high write speed by
requiring block writes. So, in order to write 10 bytes you have to
write a whole block, which may take milliseconds, but yields a high
throughput because of the size of a block (EEPROM took milliseconds to
write 1 byte, flash takes milliseconds to write several kilobytes, but
neither can complete any size write operation in less than
milliseconds). Am I wrong?

Sean


On 6/2/07, Tony Smith <RemoveMEajsmithKILLspamspamrivernet.com.au> wrote:
{Quote hidden}

> -

2007\06\03@043721 by Tony Smith

picon face
{Quote hidden}

Max writes has gotten better, but nowhere near what a HDD can do.  For
longevity, I've seen anything from 'lunchtime on Wednesday' to '27 years'
for how long a cell will last.  A million writes shows up a lot, so 1 write
per second gives 11 days before failure.  So 'lunchtime NEXT Wednesday'
then.

Journaling will increase life, and many flash cards claim to do 'wear
levelling', where the controller evens out the writes across the card.
Presumably the controller handles bad blocks as well (like HDD SMART).

One solution is just to put in more RAM so you don't need any virtual
memory.  Flash would probably work ok without that burden.

Someone once mentioned to me that UDF (for rewritable CDs) does wear
levelling too.  I've never bothered to find out if this is actually true, or
was made up by someone in marketing after lunch.

I'm not sure how flash cards handle writes, but that would a limit of the
controller, not the memory itself.

Tony

2007\06\03@100046 by Gerhard Fiedler

picon face
Sean Breheny wrote:

> Has anyone addressed the problem of max number of writes, though?

Yes. Windows XPe:
<http://msdn2.microsoft.com/en-us/embedded/aa731409.aspx>. I've also seen
distros for Linux that offer solutions for this; just don't have any links
handy.

> Most modern OSes use virtual memory extensively. Having to cache the
> flash drive to prevent memory cell wear-out due to writes defeats the
> purpose of virtual memory, since you would need almost as much cache
> memory (I think) as the virtual memory size.

Right. So you just need to work without virtual memory and do most of the
temp stuff in RAM, which means that you need enough RAM. But once you spend
the money on a Flash disk of the necessary size for a typical PC, the
additional money for the necessary RAM is little more than pocket change :)

There is also a tradeoff between having more RAM and using up more Flash
write cycles for temp stuff. This tradeoff depends on what you want or
need.

> Also, I thought that flash achieved its relatively high write speed by
> requiring block writes. So, in order to write 10 bytes you have to write
> a whole block, which may take milliseconds, but yields a high throughput
> because of the size of a block (EEPROM took milliseconds to write 1
> byte, flash takes milliseconds to write several kilobytes, but neither
> can complete any size write operation in less than milliseconds).

Correct, but with suitable caching, I don't think that any of this is
noticeable unless you're really needing high throughput writes. Harddisk
head positioning is also not instantaneous, and just as you need to write a
whole block if you want to write 10 bytes, you need to move the head to the
target location even if you only want to write 10 bytes. I didn't test it,
but I'd say for writing 10 bytes, a typical Flash is way faster than a
typical harddisk on average.

Gerhard

2007\06\03@200910 by Peter Todd

picon face
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, Jun 03, 2007 at 11:00:27AM -0300, Gerhard Fiedler wrote:
> Sean Breheny wrote:
>
> > Has anyone addressed the problem of max number of writes, though?
>
> Yes. Windows XPe:
> <http://msdn2.microsoft.com/en-us/embedded/aa731409.aspx>. I've also seen
> distros for Linux that offer solutions for this; just don't have any links
> handy.

On Linux you have a lot of options for avoiding writes to disk. It's
quite easy to disable swap entierly, you can also set swap to be only
used if you have truely run out of memory rather than the usual swap
stuff out to have more cache ram.

You can also move stuff like log files into ramdisks very easilly. They
would be deleted on power down, but for an embedded application this may
be appropriate.

You can even set the disk caching mechanism to practically never flush
writes to disk. Originally this feature was developed to extend the
battery life on laptops, but it'd be appropriate for flash as well.

Finally you can use the money you saved on software licenses to just buy
a harddrive.

> Correct, but with suitable caching, I don't think that any of this is
> noticeable unless you're really needing high throughput writes. Harddisk
> head positioning is also not instantaneous, and just as you need to write a
> whole block if you want to write 10 bytes, you need to move the head to the
> target location even if you only want to write 10 bytes. I didn't test it,
> but I'd say for writing 10 bytes, a typical Flash is way faster than a
> typical harddisk on average.

Only if you need to flush the write to flash immediately. I can't speak
for other OSes, but on Linux many of the filesystems have very advanced
write re-ordering code. So if you make 1000 tiny changes scattered
around the disk, the changes aren't flushed to disk immediately but
instead linux saves them in the cache until enough adjacent writes are
accumulated to make it worth a write to disk.

Of course for high-integrity stuff, like databases where you must have
transactions immediately written to disk, and in particular orders to
satisfy transaction constraints, then yes flash is a huge win and indeed
there are already many flash or battery-backed ram based cache layers
available for those applications.

- --
http://petertodd.ca
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGY1Yp3bMhDbI9xWQRAlvaAJ9Ov3JJvPErdbRpVuSHEY9WgFcYsgCfTtqO
gyO/UODrtB2/CQIj7/BsRLY=
=JFR5
-----END PGP SIGNATURE-----

2007\06\08@063940 by Jonathan Hallameyer

picon face
On 6/3/07, Gerhard Fiedler <listsSTOPspamspamspam_OUTconnectionbrazil.com> wrote:
{Quote hidden}

> -

2007\06\08@091402 by Gerhard Fiedler

picon face
Jonathan Hallameyer wrote:

> But once you add more ram windows, just wants even more virtual
> memory...

I'm not saying that it does work now with all systems, but I'd bet my a**
that it will be made work once flash drives enter mainstream :)

OTOH, WinXPe is somewhat different than WinXP. It might already work -- you
never know until you tried :)

Gerhard

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...