tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Import of XZ



On Wed, Aug 04, 2010 at 11:29:14AM -0400, der Mouse wrote:
> > I want to import XZ for use in pkgsrc and possible for release
> > distribution.  XZ is the successor of LZMA.  It uses the same core
> > algorithm, [...]
> 
> > LZMA as algorithm is essentially LZW with huge window size and better
> > entropy encoding.
> 
> How huge is "huge"?  I'm concerned about small-memory machines (I still
> have an hp300 with, well, "total memory = 6136 KB" / "avail memory =
> 2648 KB") finding themselves unable to even uncompress things, at all.

The default settings (auto-tuning based on available memory at
compression time) requires 10MB for the decompression. The dictionary
size can be tuned manually using the usual -0..-9 options. From various
tests it seems like 10MB dictionary size is a point of dimishing
returns.

I'm not sure if running a "modern" NetBSD without customisation on a
platform with less than 10MB of free memory is useful other than to play
around. But please keep in mind that the availability of the tools and
forcing everyone to use it are separate discussions. Keeping support for
gzip'd sets certainly has its merits due to the higher compression
speed. That is noticable for the typical developer build cycle after
all.

Joerg


Home | Main Index | Thread Index | Old Index