[Coco] CF and SD life expectancy

[Sean]

Here is a test someone did, although the article is 3 years old.

http://www.bress.net/blog/index.php?url=archives/114-How-Long-Does-a-Flash-Drive-Last.html

On Sun, Jul 24, 2011 at 6:43 PM, Roger Taylor <taylor at newfoal.com> wrote:
> At 04:31 PM 7/24/2011, you wrote:
>>
>> Interesting. Cross posted from the BBC-Micro mailing list.
>>
>> QUOTE:
>> Most of the consumer-grade CF and SD cards you can buy today are probably
>> using MLC flash which requires fairly careful supervision if you want your
>> data to stay intact.  In particular, each block of MLC can only be read from
>> a certain number of times before it needs to be erased and rewritten.
>> There's also the wear-levelling issue which is more severe with MLC because
>> the write cycle endurance of each block is less than SLC by about a factor
>> of 10.
>>
>> That's not to say that it's necessarily impossible to make a reliable card
>> out of MLC, but it requires a fairly sophisticated controller and my
>> experience has been that a lot of the lower-end cards either don't have the
>> right level of technology for that, or haven't been appropriately tested and
>> don't actually work correctly.  On some of the cards we tested recently at
>> work, just reading from the same file a couple of thousand times would cause
>> the data to gradually become corrupt.
>> /QUOTE
>>
>> Remember that NitrOS9 reads the first sector and the File Allocation
>> sectors *frequently*.
>>
>> Regards, Bob Devries
>> Dalby, QLD, Australia
>>
>> --
>> Coco mailing list
>> Coco at maltedmedia.com
>> http://five.pairlist.net/mailman/listinfo/coco
>
>
> We've dealt with flakey floppy disks since 1980 and now memory cards are far
> more reliable, yet we worry more.  So once every 3 years or so you can run a
> sector refresher over the entire card or just clone the card onto a new one.
>  They're cheaper than a bag of potato chips.
>
> Why not just test your theory with a looping BASIC program that reads LSN0
> forever until you get corrupt sector data, and post what the read count was.
>  Do the same with a write/read test to LSN0 on another card and post your
> results.  With a program running around the clock, it shouldn't take long to
> solve this mystery.  Also take note of the date/time so we can estimate the
> lifespan we can expect from typical usage on the CoCo.
>
>
>
>
>
> --
> ~ Roger Taylor
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco
>