[Coco] 2gig CF card killed (false alarm)

Roger Taylor operator at coco3.com
Sat Mar 22 16:10:15 EDT 2008


It turns out the card was corrupted somehow while I was moving it 
between my various PCs.  Vista will definately trash a FAT-16 CF, 
trust me.  It will even ask you if you "want us to fix it?", and then 
bam, won't boot on the system it was previously booting on.

Anyway, I have restored my 2gig card, so my e-mail was a false alarm 
about the write limits being reached already.  I hope not!

By the way, 300,000 cycles doesn't tell anybody much unless they know 
*exactly* how their OS dumps data to it, how often, how many cycles 
it wastes in the process, etc.





At 12:33 AM 3/22/2008, you wrote:
>Roger,
>
>I've been testing flash devices as hard drives at work for performance
>related metrics.  On a sort of related note, many apps actually run slower
>on flash drives.  They tend to write frequent, small changes to the drive in
>order to save data.  This has the two unexpected affects on flash devices.
>First, since they have to perform a block read before writing any data, they
>don't get the speed increase solid state would seemingly provide.  Second,
>any app which uses a swap file (including the OS) can very quickly burn out
>a flash device.
>
>(Vista ReadyBoost actually accounts for this and limits writes.  A typical
>swap file from Windows/Linux/Unix would not.)
>
>Does OS-9 use a swap file?  If do, is there any way to limit it writing to
>the drive?  And if it doesn't use a swap file, then I feel pretty
>comfortable saying the flash device would outlast the Coco itself.
>
>John Guin
>
>-----Original Message-----
>From: coco-bounces at maltedmedia.com [mailto:coco-bounces at maltedmedia.com] On
>Behalf Of Roger Taylor
>Sent: Friday, March 21, 2008 10:05 PM
>To: CoCoList for Color Computer Enthusiasts
>Subject: [Coco] 2gig CF card killed
>
>Well, I think I've successfully killed a nice Lexar 2gig CF card after many
>attempts of installing Windows 98SE on it to run on my Compaq IA-1 internet
>appliance.
>
>The IA-1 was hacked through software by replacing the internal 16mb sandisk
>(originally stocked with MSN Companion, a browser system) with Midori Linux.
>Since the unit can also boot from the CF slot, a FAT-16 formatted card made
>bootable and with MS-DOS system files would boot into MS-DOS, and if Windows
>98 was installed on the FAT-16 card, it would boot as well.
>
>I was trying different install methods, first putting the Win98SE CD
>contents on the MS-DOS bootable CF card, then running setup.exe from the DOS
>prompt on the IA-1.  This worked perfect up until it kept locking up far
>into the install when the plug and play detection was happening.  Then I
>installed WIndows on the CF from my PC with the CF card connected as IDE
>drive 0, primary, using an IDE to CF adaptor.  This worked great and Windows
>and the PC both thought it was a real drive.
>
>I did so many installs and formats, that I think I reached the ~300,000
>erase/write limit of the card.
>
>My question is, with the IDE interfaces in use and people using CF cards as
>their main CoCo HD, how long would you expect the card to make it as a hard
>drive knowing that the cards were designed with a limited number of writes
>possible, and also when the card reaches this point, is it readable-only
>then?  Mine can't even be accessed now.
>
>It seems to me that more and more people are trying to use CF cards as hard
>drive solutions for embedded systems and even for their computers.  This has
>got to be the business to get into?  :)  Think about it, they've designed a
>card that really shouldn't be any different than a memory stick in what they
>do (store memory and read it), but for some reason the CF's have a dying day
>somewhere in the future, sooner or later, depending on your use.  They know
>very well that people are trying to use them as hard drives on various
>systems, and that unless it's an embedded solution like Windows has done
>with a version of CE to limit the # of commits to the card, it's a dead card
>the day you buy it.  I don't think they're worth messing with.
>
>
>
>--
>Coco mailing list
>Coco at maltedmedia.com
>http://five.pairlist.net/mailman/listinfo/coco





More information about the Coco mailing list