[Beowulf] MS HPC... Oh dear...

Ashley Pittman ashley at quadrics.com
Mon Jun 12 09:20:41 PDT 2006


On Mon, 2006-06-12 at 11:56 -0400, Joe Landman wrote:

> > technical problems of making a stable, cross ISV ABI are the same on
> > Linux and Windows (possibly even slightly worse under Windows), the only
> > difference being that under Linux there is a large amount of inertia in
> > keeping the status quo as moving to a consistent model would break at
> > least somebodies applications, under a brave new Windows world that
> 
> Which is fine (breaking of the application), as the benefits IMO far
> outweigh the costs.  Simple CBA:
> 
> Cost:
> 	some apps will break
> 	some will need to be recompiled, with some minor source code adjustments
> 
> Benefits:
> 	will work with new hardware drivers compiled as .so/.dll linked into
> the base (LAM, Scali, and others demonstrate that you can do this).

The point I was trying to make is this cost reduces to zero when you
move to a new architecture.  If ever a ABI is going to be adopted that
is the time to do it...

Whilst it may be an accidental benefit of Windows it's hardly a
compelling argument for doing so.

> My current belief is that Microsoft will create/mandate the model by
> supplying the MPI.  Some standards are defined after the fact.
> 
> I am not advocating mimicing the Microsoft ABI.  I am advocating getting
> a single MPI ABI per ISA ABI.  The question of course is, which one.

Aren't these two statements contradictory? 

Ashley,




More information about the Beowulf mailing list