sram vs. flash - which is better?
Mark Ackerman
faster, but eats more battery power. The flash takes power only
on writing.
Is this correct? I'd like to double-check my understanding before
going out and buying a card.
There was recently a post here stating that the sram doesn't consume
main battery power, only the battery on the card. If this is true,
and the battery lasts a year, it would seem this isn't much of a
concern. Why bother with flash then? It seems more expensive,
and according to Tiger Direct, takes 10 milliseconds on access.
Is there a problem with replacing batteries on sram?
Any opinions on which is better for normal use?
Mark Ackerman
("Ack")
Information and Computer Science
University of California, Irvine
BTW, thanks to Anthony Stieber for his great tutorial on PCMCIA cards.
Albert Nurick
: Any opinions on which is better for normal use?
IMO, both work well. It's based on the capacity you need. If it's 2MB or
less, stick with SRAM. If it's 3MB or more, consider flash.
I've got a 10MB flash card in my 100LX, and it's *great*.
: BTW, thanks to Anthony Stieber for his great tutorial on PCMCIA cards.
Absolutely. Anthony does an outstanding job; he has perhaps the best
signal-to-noise ratio of anyone on Usenet.
--
| Albert Nurick | "In case of doubt, decide in favor of |
| alb...@bga.com | what is correct." |
| http://www.realtime.net/~albert | -- Karl Kraus |
Alan K Biocca
Mark Ackerman <acke...@ics.uci.edu> wrote:
>If I understand correctly, the difference is largely that sram is
>faster, but eats more battery power. The flash takes power only
>on writing.
The flash requires power when reading or writing but does not require
power to hold data.
>Is this correct? I'd like to double-check my understanding before
>going out and buying a card.
>There was recently a post here stating that the sram doesn't consume
>main battery power, only the battery on the card. If this is true,
>and the battery lasts a year, it would seem this isn't much of a
>concern. Why bother with flash then? It seems more expensive,
>and according to Tiger Direct, takes 10 milliseconds on access.
>Is there a problem with replacing batteries on sram?
SRAM is powered by the HPLX when the HPLX is on, otherwise it uses
its own backup power. To change the battery in SRAM put into HPLX
and keep HPLX turned on while removing and replacing SRAM battery.
Make sure the HPLX doesn't time out and auto-shutoff during this
operation!!!
Flash is considerably cheaper per byte than SRAM at normal prices.
Flash has considerably more capacity than SRAM per card.
Flash is somewhat more robust than SRAM since a shock can cause
battery backup power to be interrupted. There is little warning that
the backup battery is low. Occasionally batteries go bad early.
>Any opinions on which is better for normal use?
Depends on what 'normal' use is. If you want to carry lots of read-mostly
data the Flash is supreme. If you want to constantly rewrite much of
your data then SRAM has an advantage. If you only want 1 meg then
cheap newton SRAMs are hard to beat. If you want more than 2 megs then
flash is it again.
I started with a 1 meg SRAM and it worked fine. I later got 10 meg flash
and am much happier. I keep my dynamic files in the built-in ram and
move them out to the flash when they are less dynamic. I make backups
of the whole HPLX ramdisk using pkzip periodically and keep a rotating
set of them on the flash.
I recommend getting a cheap 1meg SRAM to start with and see how that
works out. A flash is a bigger investment and it is important to
decide on the size you want before making that purchase.
Alan K Biocca
Steve Collins
> acke...@ics.uci.edu (Mark Ackerman (acke...@ics.uci.edu)) wrote:
> : Any opinions on which is better for normal use?
>
> IMO, both work well. It's based on the capacity you need. If it's 2MB or
> less, stick with SRAM. If it's 3MB or more, consider flash.
>
> I've got a 10MB flash card in my 100LX, and it's *great*.
>
> : BTW, thanks to Anthony Stieber for his great tutorial on PCMCIA cards.
As I understand it (which sadly isn't very well) the flash card needs to
be erased before being written to. It's this double proceedure that slows
them down. I don't know if that is correct. If it is what bothers me
is that they may wear out (in the same way as a cassette tape wears out if
you continually record and erae it). Is this true or ?
--
Steve Collins Concern 248-250 Lavender Hill, London, SW11 1LJ.
Tel: 071 738 1033 Fax: 071 738 1032 E-Mail: st...@concern1.demon.co.uk
Albert Nurick
: As I understand it (which sadly isn't very well) the flash card needs to
: be erased before being written to. It's this double proceedure that slows
: them down.
That's the theory. In my experience, Flash performs very well.
: I don't know if that is correct. If it is what bothers me
: is that they may wear out (in the same way as a cassette tape wears out if
: you continually record and erae it). Is this true or ?
Engineers I've spoken with report that flash memory's "wearing out" isn't an
issue if you'll be using 'em as a disk drive; I've heard numbers like
"twenty years" tossed around, assuming typical usage.
Nigel Orr
Steve Collins <st...@concern1.demon.co.uk> wrote:
>
>As I understand it (which sadly isn't very well) the flash card needs to
>be erased before being written to. It's this double proceedure that slows
>them down. I don't know if that is correct. If it is what bothers me
>is that they may wear out (in the same way as a cassette tape wears out if
>you continually record and erae it). Is this true or ?
From my sketchy recollection of MOS device theory... it shouldn't have
any effect on lifetime- ie the lifetime of a Flash device should be no
shorter than any other semiconductor. The "erase before read" bit
isn't really correct either, as I recall- I can't remember what makes
it slower, but I would suspect that it is simply higher gate
capacitance.
It certainly bears very little, if any, resemblance to the process of
transferring data on mag tape...
Nigel
--
Nigel o...@tks.oulu.fi
Daniel T Senie
>St...@concern1.demon.co.uk (Steve Collins (St...@concern1.demon.co.uk)) wrote:
>: As I understand it (which sadly isn't very well) the flash card needs to
>: be erased before being written to. It's this double proceedure that slows
>: them down.
>
>That's the theory. In my experience, Flash performs very well.
>
>: I don't know if that is correct. If it is what bothers me
>: is that they may wear out (in the same way as a cassette tape wears out if
>: you continually record and erae it). Is this true or ?
>
>Engineers I've spoken with report that flash memory's "wearing out" isn't an
>issue if you'll be using 'em as a disk drive; I've heard numbers like
>"twenty years" tossed around, assuming typical usage.
The numbers I've seen lately are in the 100,000 and up erasure cycles
region. Not something to be too worried about.
Flash chips DO have to be erased in blocks before writing to them. The block
sizes vary depending on the chips used, and the layout of those chips in
a circuit (i.e. do you use 1, 2, or 4 of the chips wide depending on your
bus size, dynamic bus sizer, etc.), and on the bit width of your flash
chips (many are 8 bits wide, some are 16, etc.).
Some of the earlier flash parts, notably early 28F010s had some failure
modes in which software could permanently destroy the chips. Newer chips
don't have that problem.
--
---------------------------------------------------------------
Daniel Senie Internet: d...@world.std.com
Daniel Senie Consulting n1...@world.std.com
508-779-0439 Compuserve: 74176,1347
Jerry Fountain
Daniel T Senie <d...@world.std.com> wrote:
>In article <35lj23$5...@giga.bga.com>, Albert Nurick <alb...@bga.com> wrote:
>>St...@concern1.demon.co.uk (Steve Collins (St...@concern1.demon.co.uk)) wrote:
>>: As I understand it (which sadly isn't very well) the flash card needs to
>>: be erased before being written to. It's this double proceedure that slows
>>: them down.
>>
>>That's the theory. In my experience, Flash performs very well.
>>
>>: I don't know if that is correct. If it is what bothers me
>>: is that they may wear out (in the same way as a cassette tape wears out if
>>: you continually record and erae it). Is this true or ?
>>
>>Engineers I've spoken with report that flash memory's "wearing out" isn't an
>>issue if you'll be using 'em as a disk drive; I've heard numbers like
>>"twenty years" tossed around, assuming typical usage.
>
>The numbers I've seen lately are in the 100,000 and up erasure cycles
>region. Not something to be too worried about.
This sounds reasonable, until you think of how a DOS type file system will
operate. While a typical file area might not be a problem, consider the FAT
area. This region will undergo the most frequent R/W activity.
Consider a "worst case" of using flash memory as your TEMP storage. Consider
a program like Windows running using temp to store print files, applications
using it for timed backups, etc. You can easily envision 100-200 file
creations/deletions per day. Each creation/deletion is at least TWO writes,
times say 100 per user day, 5 days per week, 50 weeks per year (allow for
some vacation) = 50,000 writes/year. Thus this *region* could easily fail
within two years. I'm not saying it will, but one must consider the area
most prone to heavy R/W activity.
Jerry
--
Jerry Fountain | Laboratory for Fluid Mechanics, Chaos, and Mixing
g...@chem-eng.nwu.edu | Northwestern University
(708) 491-3555 (Office) | Department of Chemical Engineering
(708) 491-3728 (FAX) | 2145 Sheridan Road, Evanston, IL 60208
Joel Spolsky
>
>This sounds reasonable, until you think of how a DOS type file system will
>operate. While a typical file area might not be a problem, consider the FAT
>area. This region will undergo the most frequent R/W activity.
>
Microsoft's file system for Flash is not exactly FAT. It is a
derivative of FAT that is designed to spread out the changes and
writes as evenly as possible, over the long term, over the surface of
the flash disk. So you won't "wear out" your flash disk by burning a
hole in one spot on the disk, the operating system will cleverly
spread out things as much as possible.
Joel
--
Joel Spolsky
spo...@panix.com
Charles Ader
: As I understand it (which sadly isn't very well) the flash card needs to
: be erased before being written to. It's this double proceedure that slows
: them down. I don't know if that is correct. If it is what bothers me
: is that they may wear out (in the same way as a cassette tape wears out if
: you continually record and erae it). Is this true or ?
Flash memory as I understand it is SRAM plus EEROM in each bit cell. This
means that the whole array can be written in one EEROM write cycle. Power
consumption restraints may mean that blocks of memory must be written over
several write cycles. In any case the Flash memory is supposed to written
only when you turn off power to the PCMCIA card. Most EEROMs work well
past 10,000 write cycles and the newest ones last beyond 100,000 cycles.
In general the power button should wear out before the Flas memory.
Charles.