Why Do Flash Drives Cost So Much?
Alvin Pettit asks: "I wanted to get a Flash drive for my PC for the following reasons:
it is quiet, I can save electricity and I don't have to worry about moving parts. When I looked for these drives I found them to be rather expensive, much more so than the smaller devices such as CompactFlash! Why do Flash drives cost so much more than CompactFlash devices?"
"I looked up IDE flash drives compared to compact flash and this is what I found:
- On pricegrabber:
SanDisk Part# SD25B880402 880MB IDE 2.5 FLASHDRIVE is $1148.00
This comes out to about $1.30 per meg - Where a compact flash is
SanDisk Part# SDCFB1000768 1 GB COMPACTFLASH CARD is $589.00
This comes out to about $.60 per meg - Even Ultra Compact flash is cheaper:
SanDisk Part# SDCFH512784 512MB COMPACT FLASH ULTRA is $268.00
This comes out to about $.52 per meg
a 1 gig flash card is limited by your USB speeds
a 1 gig flash drive is limited by your system bus speeds
the drives are MUCH faster
Buttsex.
From what I've seen , and what the guys at wearables it is indeed possible to construct a low power pc that boots off a PCMCIA (adapted CF) card.
Although their end goals are not identical as yours, their immediate needs (low power) are the same.
Why does nobody make one with a write-protect switch???
They would be perfect for storing Tripwire databases, read-only boot partitions, etc. I've looked all over, though, and as far as I can tell, all of them are permanently read/write.
They're not the same thing, or at least they shouldn't be. Flash memory is *really* slow, fast random access, but spectacularly slow read/write. And it wears out. A good quality flash drive should be a stack of DRAM, a battery, and some way of backing up the DRAM when the power gets yanked (and vice versa). As you can imagine, this costs a bit and since it has low demand, it is also expensive.
Dave
I write a blog now, you should be afraid.
Why do flash drives cost so much more? Most likely because they aren't easily found. They're not used much, and I'd assume that most of them have very fast access times (which is what you're paying the most for. Faster chips can be expensive as hell, but I bet there is nothing like being able to saturate your IDE channel with just one drive that you can't even hear). Of course this doesn't make a ton of sense, because to put a gig in a little CF card, the chips have to be incredibly small and dense. To put a gig of memory into something the size of a hard drive wouldn't need very dense or small chips (relitivly) and they could use more chips of lower densities so they should be able to get a decent discount.
My last comment for you is this: the ATA specification is very well documented, and RAM is cheap. If people can interface PIC chips, HC11s, FPGAs, and other things to IDE, they someone could too. I wouldn't be terribly suprised if there was a project out there somewhere (shouldn't be TOO hard to do anyway) to basically turn a bunch of RAM into an IDE drive. Then all you'd need is some sort of battery to keep it going when the PC is off. Plus it'd be easily upgradeable.
Comment forecast: Bits of genius surrounded by a sea of mediocrity.
Why do you want them more?
You do NOT want to use a compact flash card for a read/write file system; they have a limited number of write cycles.
You're thinking of a RAM drive. These usually present a SCSI interface, and are really horrendously expensive. Often used to accelerate database performance on mid-range ($100K) solaris servers.
There are a number of companies selling actual "flash" drives, both as CF-to-IDE harnesses and custom packaged in a laptop-drive form factor.
These are nothing like RAM drives, and in fact are not really any more sophisticated than your standard "Compact Flash" storage card.
Here's an example with some specs:c hip.htm
http://www.acal.be/products/el/active/sandisk/san
I have a couple of 64Mb models, you can often find them on Ebay at reasonable prices. I use them to build Diskless FreeBSD hosts.
I do not deploy Linux. Ever.
Flash memory can only be written to approx. 100,000 times in any one spot before it will fail. Flash drives (and compact flash) will try to distribute the load, but if you have anything running that is caching to the drive it can wear out quickly. So things like the tmp directory should go in RAM.
USB has nothing to do with CompactFlash...
I have a CF->PCMCIA adapter - MUCH faster than a USB reader, and in fact indistinguishable from the much more expensive ATA PCMCIA cards.
CF cards have a built-in IDE interface, connecting them to an IDE bus is a matter of passive wiring. (There are adapters to do this for $10-20, MAYBE $30, but I'm positive it's not more than that. My CF-PCMCIA adapter cost me $10)
retrorocket.o not found, launch anyway?
Are you trying to use "solid state" as a synonym for "non volatile"?
Two reasons- power and refresh- back when 4M sticks were common- the ram was pretty power hungry, if i remember right, it was about 1W/MB (at 5V). Even if that estimate is really high, it would still take a lot of power to keep up a whole drive worth of DRAM. The second reason is that the "D" in DRAM stands for dynamic- it is really just a single capacitor that drains off rather quickly- on the order of milliseconds, generally. You need to have something read the memory cell, and rewrite it. You really want SRAM- or static ram, but that takes 4 transistors/bit of memory, so they tend to be much less space efficient (and more expensive) than DRAM (which only needs 1 transistor/bit). SRAM doesn't need refreshing, and in general, as long as you aren't reading or writing to it, the current draw is really low.
I am going to guess that a momentary loss of attention caused c0ldfusion to write "solid state" memory when he or she meant "static" memory.