[gnutls-dev]Re: LZO algorithm

Nikos Mavroyanopoulos nmav@gnutls.org
Tue Sep 17 22:55:01 2002


On Tue, Sep 17, 2002 at 09:21:22PM +0200, Simon Josefsson wrote:

> HTML FILE (270kb):
> ZLIB: Data transferred:          126267495 bytes
> LZO:  Data transferred:          179489923 bytes
> NULL: Data transferred:          172701348 bytes

> Are these figures accurate?  LZO results in larger than NULL?  And
> none of the figures seems close to 270KB?

If I haven't done any mistake, these figures should be accurate. 
The 270kb file was downloaded several times, by a pool of 15
clients. The web server used was boa, and the test program was siege
(both of them modified for gnutls).

LZO -- and a compression algorithm in general -- can be faster than null,
if there is actual compression, and the time spend in compression is a small
fragment of the time needed for the MAC and encryption. It seems that
LZO is really fast :)

> Thanks.

-- 
Nikos Mavroyanopoulos
mailto:nmav@gnutls.org