James Kehl wrote:
> Just to confirm, you're talking about doing a partial page read of the
> *main* page area, not the spare? i.e. only reading out the first 64
> bytes or so of a page, not bothering with any of the standard ECC and
> just relying on the separate tags ECC?

Correct. Provided the chip supports random reads (most do, I believe), 
your in-tags ECC is sufficiently complex for the NAND part (multiple 
plane chips could make life interesting - check the data sheet for 
recommendations), and you take care to not send any hardware ECC unit 
into conniptions, I think you're looking at a worthwhile speed-up, 
especially on boards with a relatively slow clock speed. The downside, 
of course, is that you lose however many bytes of every page to the tags 
ECC as well as the tags themselves.


Ross

-- 
eCosCentric Ltd, Barnwell House, Barnwell Drive, Cambridge CB5 8UU, UK
Registered in England no. 4422071.                 www.ecoscentric.com