On 10/5/06,
ian@brightstareng.com <
ian@brightstareng.com> wrote:
> There is a lot of value knowing the underlying performance of
> NAND i/o on a platform and I agree with you that there is value
> in comparing real NAND with nandsim and seeing what happens to
> the filesystem performance -- what I didn't like about the
> original posting was the lack of baseline facts like NAND i/o
> rates/timing, nandsim parameters etc., without these I can't
> deduce very much from the figures.
Well, if I find time to do a more comprehensive comparison, I'll do.
As of now, I just collected a sort of preliminary results which
weren't expected by me so I did post that. Consider it a food for
thought, basically :)
I don't think that the chip delays matter much here, but anyway YES,
the controller and the whole situation is a bit uncommon here. First
of all, the NAND controller on PNX4008 is DMA-capable and all the
testing was performed in DMA mode. Then, it has HW ECC capabilities
and imply a bit weird layout of the data vs OOB within the page, so in
order to write OOB you have to issue RND commands.
Given all that, I do encourage anyone who cares to do similar or
better measurements on some other 2k page NAND so we could compare the
results :) But my view on the results is that we should start moving
the tags inband for 2k page flashes...
Best regards,
Vitaly
(text/plain)