Mailing List Archive

Performance testing
Hi,

Does anyone have ideas on how to do a performance test of the various
cipher suites? I'm asking because I would like to see how much speed
increase can be gained by compiling optimized libraries for the Debian
package.


Ivo

--
Als auch die vierten Kinder einfach vorbeilaufen, beschlusst die Hexe,
das Haus aus Schafskäse aufzugeben und nach einer Immobilie aus
Lebkuchen ausschau zu halten.
- Nichtlustig
Re: Performance testing [ In reply to ]
On Thu Nov 21 2002; 21:53, Ivo Timmermans wrote:

> Does anyone have ideas on how to do a performance test of the various
> cipher suites? I'm asking because I would like to see how much speed
> increase can be gained by compiling optimized libraries for the Debian

What about the benchmark program that comes along with libgcrypt?


Timo
Re: Performance testing [ In reply to ]
On Thu, Nov 21, 2002 at 09:53:35PM +0100, Ivo Timmermans wrote:

> Hi,
> Does anyone have ideas on how to do a performance test of the various
> cipher suites? I'm asking because I would like to see how much speed
> increase can be gained by compiling optimized libraries for the Debian
> package.

I've done some testing using a gnutls http server as a benchmark, and
some polling http clients. Using the -march and -mcpu options for
my cpu (pentium3), I've noticed an increase in speed, of about 30%.


> Ivo
> --
> Als auch die vierten Kinder einfach vorbeilaufen, beschlusst die Hexe,
> das Haus aus Schafsk??se aufzugeben und nach einer Immobilie aus
> Lebkuchen ausschau zu halten.
> - Nichtlustig

--
Nikos Mavroyanopoulos
Re: Performance testing [ In reply to ]
Hi Ivo,

can you please give us a short overview on the architecture specific
packaging things? I noticed that the runtime linker first tries a
subdirectories of /usr/lib to get a CPU optimized version. Is that
scheme somewhere described and do you know whether there is some
support in libtool planned?

I have tought for quite some time on how to do CPU specific
optimizations for the assembler code. It is tricky because we can't
simply use a switch or a function pointer due to the extra overhead.
The only solution within libgcrypt would be to duplicate larger parts
of the code - something I am not keen to do: It increases complexity,
introduces errors and we have to find a portable way to figure out the
best CPU to use.

Building extra libraries or splitting libgcrypt into a general and a
performance critical part seems to be a far better idea.


Shalom-Salam,

Werner
Re: Performance testing [ In reply to ]
Werner Koch wrote:
> Hi Ivo,
>
> can you please give us a short overview on the architecture specific
> packaging things? I noticed that the runtime linker first tries a
> subdirectories of /usr/lib to get a CPU optimized version. Is that
> scheme somewhere described and do you know whether there is some
> support in libtool planned?

I'm afraid I don't have an answer to these questions yet. My plan was
to steal some things from the Debian package for OpenSSL.

> Building extra libraries or splitting libgcrypt into a general and a
> performance critical part seems to be a far better idea.

I agree.


Ivo

--
Um den Elefanten durch den Zoll zu schmuggeln, hatte Rasmus ihn als
Schäferhund verkleidet.
- Nichtlustig
Re: Performance testing [ In reply to ]
Timo Schulz wrote:
> On Thu Nov 21 2002; 21:53, Ivo Timmermans wrote:
>
> > Does anyone have ideas on how to do a performance test of the various
> > cipher suites? I'm asking because I would like to see how much speed
> > increase can be gained by compiling optimized libraries for the Debian
>
> What about the benchmark program that comes along with libgcrypt?

D'oh! I feel silly. :)


Ivo

--
"C combines the power of assembler with the portability of assembler."
- Anonymous, alluding to Bill Thacker