First, the SanDisk documents are not consistent in terminology. In any event, consider a block a group of 512 byte sectors. A write is always a block, no matter how many sectors you wrote. The block size varies card the card but can be read in. The card has enough ram to buffer an entire block, and you can do a multiple sector write. If you can make the driver write more than one sector, you could considerably extend the life. If you wrote just two sectors, you would double the life and so on.
Also note that the max write time is 250ms, probably due to the shuffling. Writing multiple sectors would greatly increase the write speed. Also, as the file system driver level, caching the FAT entries would also make a big difference. 2009/3/26 Bobby Clark <[email protected]> > Hi Jim, > > Thanks for sending that out. After reading page 3 it looks like the card > maintains a pool of available addresses per zone and uses a new one for each > new write. When the available pool of 3 % per zone is used up the card will > fail. The wear leveling is not over the entire card as of October 2003 the > date of the document. If this is true for a current card the OS would have > a direct influence on life if it used the same zones for data writes. Most > of the current flash are the multiple bits per cell. This may have a much > more robust scheme. I have not been able to wear out any SD card yet, but > all of mine are made in the last 2 years. We should be able to make a > simple wear out tester to find out what the real limits are on specific > cards. This testing would be limited to cards of the same lot. > > Bobby > > > ----- Original Message ----- > *From:* Jim Donelson <[email protected]> > *To:* uClinux development list <[email protected]> > *Sent:* Thursday, March 26, 2009 5:27 AM > *Subject:* Re: [uClinux-dev] SD card corruption upon reboot and de-power > > : > >> Jim Donelson wrote: >> >>> -If we are talking about regular SD cards, there is no internal wear >>> leveling. >>> It is up to the driver to do that. >>> >> >> I was wrong about that, there is internal wear leveling. > Read this: > > http://www.sandisk.com/assets/file/oem/whitepapersandbrochures/rs-mmc/wpaperwearlevelv1.0.pdf > > I also reviewed the interface spec, and there is not documented way to > reset it. > There seems to be much hand waving here too about how it actually does > this, and what would happen if you remove power while these operations are > in progress. > > ------------------------------ > > _______________________________________________ > uClinux-dev mailing list > [email protected] > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by [email protected] > To unsubscribe see: > http://mailman.uclinux.org/mailman/options/uclinux-dev > > > _______________________________________________ > uClinux-dev mailing list > [email protected] > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by [email protected] > To unsubscribe see: > http://mailman.uclinux.org/mailman/options/uclinux-dev >
_______________________________________________ uClinux-dev mailing list [email protected] http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by [email protected] To unsubscribe see: http://mailman.uclinux.org/mailman/options/uclinux-dev
