Re: [RFC] LZO de/compression support - take 4

[Daniel Hazelton <dhazelton@enter.net>]

On Monday 28 May 2007 05:46:59 Nitin Gupta wrote:
> On 5/28/07, Daniel Hazelton <dhazelton@enter.net> wrote:
> > On Monday 28 May 2007 05:08:54 Nitin Gupta wrote:
> > > On 5/28/07, Daniel Hazelton <dhazelton@enter.net> wrote:
> > > > On Monday 28 May 2007 04:37:04 Nitin Gupta wrote:
> > > > > On 5/28/07, Daniel Hazelton <dhazelton@enter.net> wrote:
> >
> > New testbed based on minilzo complete.
> > Results from run using 1000 runs to generate averages:
> > 1000 run averages:
> > 'Tiny LZO':
> >         Combined: 55.196 usec
> >         Compression: 37.132 usec
> >         Decompression: 18.064 usec
> > 'miniLZO':
> >         Combined: 55.785 usec
> >         Compression: 40.862 usec
> >         Decompression: 14.923 usec
>
> Great!
>
> I believe its now ready for mainline. We can do further cleanups and
> optimizations there - more users of this code will surely drive more
> enhancements.

Forgot to mention - this is still 'take 4' - I'm going to import the 'take 5' 
code now and see if there is any appreciable difference
>
> > (using (tiny/full)/100 for percentages:
> > 'Tiny' is 0.9% faster on average
>
> I think this is more appropriate: [(full-tiny)/full]*100
> => tiny is ~1% faster on average.

Okay, I'll use that for all future calculation.

DRH

(Next version of the testbed will also test against the _safe variant in 
miniLZO)