Re: matmul intrinsic procedure

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Jun 30, 2005, at 4:40 PM, Jan Stanisław Owoc wrote:

> > P.S. Is it really not yet implemented in gfortran? That sort of
> > surprises me, but I haven't actually checked.
>
> I have been able to compile and run this program with ifort, g95 on 
> Linux
> and Cygwin, and f95 on Solaris. It uses a very simple matrix
> multiplication. It compiles, but does not run:
>
> ./faim ../dat/freetest
> faim: ../../../libgfortran/generated/matmul_r8.c:170: matmul_r8: 
> Assertion
> `a->base == 0' failed.
> Abort
>
> I can reproduce this error regardless of what kind of stuff I pass as
> arguments. This exact error has been present for at least four months 
> (in
> pre-releases of gfortran), but I was hoping it would be fixed in 4.0. 
> As
> it is not, I am interested in helping to get it working by 4.1.

It can be still fixed for a 4.0.x release, if we get the source for the 
test
you are trying.

I know that matmul is used in SPEC and that works correctly.

Please file a bug report or at least send the source to the list.

Thanks,
Andrew Pinski