extend the perfbase-0.9.0/examples/auto_compare/PMB example

[Åsmund Østvold]

As you have understood I have done some extensions to the PMB example.  Do  
you want to get a update from me on the perfbase/examples/auto_compare/PMB  
with the extensions that I have made?

I see the problem of giving real data that but we could keep  MPI-A,B and  
C in addition to OpenMPI, MPICH and so on and with made up version  
numbers.  I am guessing it is not only we me that like to be able to  
compare MPI implementations easy and in an analytic way.

Åsmund

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...

[Joachim Worringen-3]

Åsmund Østvold schrieb:
> As you have understood I have done some extensions to the PMB example.  
> Do you want to get a update from me on the
> perfbase/examples/auto_compare/PMB with the extensions that I have made?

That's a good idea. However, I intentionally simplified the examples
(removed numerous parameters) from my real-world working experiments to
make it easier to understand what going on (more compact XML files). And
because everyone typically has his own requirements for the parameters
that should be included, I think it does not make much sense to add more
parameters. Instead, users will add their own parameters as you did.

On the other hand, if you have new or modified queries or other scripts
that improve the analysis, or feel that you added a really cool way to
process input files, please send it to me.

> I see the problem of giving real data that but we could keep  MPI-A,B
> and C in addition to OpenMPI, MPICH and so on and with made up version
> numbers.  I am guessing it is not only we me that like to be able to
> compare MPI implementations easy and in an analytic way.

A, B and C in the example *are* actually OpenMPI, MPICH, and a vendor
MPI... (not in this order!). More data probably is not required to make
the example understandable, or is it?

  Joachim

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...

[Åsmund Østvold]

On Thu, 08 Jun 2006 23:04:02 +0200, Joachim Worringen <joachim@...>  
wrote:

> That's a good idea. However, I intentionally simplified the examples  
> (removed numerous parameters) from my real-world working experiments to  
> make it easier to understand what going on (more compact XML files). And  
> because everyone typically has his own requirements for the parameters  
> that should be included, I think it does not make much sense to add more  
> parameters. Instead, users will add their own parameters as you did.

I see your point.

> On the other hand, if you have new or modified queries or other scripts  
> that improve the analysis, or feel that you added a really cool way to  
> process input files, please send it to me.

Hopefully I will have something later on.  I will send it to you so you  
can look at it and use it if you find it interesting.

> A, B and C in the example *are* actually OpenMPI, MPICH, and a vendor  
> MPI... (not in this order!). More data probably is not required to make  
> the example understandable, or is it?

I understood that it was real data.  My point was that eaven if the  
example did not have input filese for other MPI implementations it would  
be nice to have them in the XML files.

But this is the way I like to have it.  And I know othes that like the  
examples to be small and only the nessesary.

I am still very happy with perfbase

  Åsmund

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...