Subject: Re: Updated libssh2 benchmark results

Re: Updated libssh2 benchmark results

From: Peter Stuge <>
Date: Thu, 18 Nov 2010 20:19:10 +0100

Mark Riordan wrote:
> > the purpose is not to significantly improve the numbers, but to
> > create an easily repeatable environment for further testing and
> > development.
> I have plenty of RAM in my systems, and I think the various
> OSes do a very good job of caching recently-used files

I would like to completely eliminate all outside factors in order to
have less disinformation to care about and compensate for when
working on this.

Mark Riordan wrote:
> Well, I'm looking for real-world performance, not just high numbers
> in my tests.

I specifically explained that this is *not* the purpose of creating
an artificial environment for testing and for development.

The goal is obviously real-world performance, but getting there will
be easier if as much noise as is possible can be removed from
measurements and data during testing analysis and development.

There is obviously a big problem somewhere, which may be easy to
find, but it will be easier to also spot other problems with hard
reliable reproducible numbers.

> Since other SSH clients are able to achieve higher
> numbers consistently on the same client and server hardware,
> I do not suspect filesystem performance as a factor here.

I'll try to make the point again; the purpose of a controlled,
artificial environment for testing and development is not to make
numbers look better, but to have a common reference for everyone, and
to have absolutely minimal interference with timing within the

Received on 2010-11-18