On Sun, Feb 04, 2007 at 11:12:50PM -0800, shailesh lomate wrote:
> I am using the JFFS2 file system in my project which
> is using a flash (having access time of 70 ns) as
> jffs2 flash.
> My project included a ARM7 processor running at 73
> MHz.
> I am using the JFFS2 file system with zlib compression
> support. My buffering for File IO is 256 bytes. Also
> the file read operation works fine but the problem
> which I am facing is the time taken by the fread call
> to jffs2 file system takes almost 18 ms to read a
> small chunk of just 256 bytes.
> Don?t this sound to be very in-efficient? Or is there
> anything wrong with my system?
I got much better sequential access by using bigger buffers. Try using
4Kbyte buffers and do 4Kbyte reads. Also make sure zlib has plenty of
memory available to it. Plus you might want to see how much of a
performance gain you get without zlib. When you have a slow CPU and
fast flash, compression could be the bottleneck. This is especially
true for writing, when the jffs2 code will try many different
compressions schemes to find which is the best for a given block of
data.
Andrew
--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
