Re: [PATCH 3/5] Add support for LZO-compressed kernels

[Albin Tonnerre]

On Wed, Jul 22, 2009 at 03:28:09PM +0000, kevin.granade@xxxxxxxxx wrote :
> On Jul 22, 2009 9:01am, Albin Tonnerre <albin.tonnerre@xxxxxxxxxxxxxxxxxx>
> wrote:
> > This is the first part of the lzo patch
> >
> > The lzo compressor is worse than gzip at compression, but faster at
> >
> > extraction. Here are some figures for an ARM board I'm working on:
> >
> >
> >
> > Uncompressed size: 3.24Mo
> >
> > gzip 1.61Mo 0.72s
> >
> > lzo 1.75Mo 0.48s
> >
> >
> >
> > So for a compression ratio that is still relatively close to gzip, it's
> >
> > much faster to extract, at least in that case.
> 
> Is that "time to run the extraction algorithm", or "time to read in image from
> media and extract"? I think the time to read from the media would tend to
> dominate the decompression time.
> Either way, could you provide the other time for each algorithm in order to
> give a sense of how this might scale to other CPU speeds/media read speeds?
> 

That's the time to run the extraction algorithm. As H. Peter Anvin pointed out,
you can have a fast media and a somewhat slow CPU, for which lzo makes sense.
As for other data, I don't have all the figures handy, but:

 - LZMA: compressed size 1.19Mo, decompression time: several *seconds* (that's
   on a 180MHz ARM9 board, using a patch to implement LZMA compression similar
   to this one)
 - Bzip2 eats a lot of RAM, and head.S only gives 64Ko of malloc() space on
   ARM, so I didn't try.

Regards,
-- 
Albin Tonnerre, Free Electrons
Kernel, drivers and embedded Linux development,
consulting, training and support.
http://free-electrons.com

Attachment: signature.asc
Description: Digital signature