help-gsl
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Help-gsl] GotoBLAS2 (was Re: Building gsl-1.15 under MinGW)


From: John Chludzinski
Subject: Re: [Help-gsl] GotoBLAS2 (was Re: Building gsl-1.15 under MinGW)
Date: Tue, 23 Aug 2011 04:24:22 -0400

First, I assume you had little or no trouble building LAPACK
using GotoBLAS2 ?

Second, I checked using:

nm libgoto2_barcelonap-r1.13.lib | grep zlange
        U _zlange_
        U _zlange_
0000000 T _zlange_
        U _zlange_
        U _zlange_
        U _zlange_

but didn't see an zlange.obj.

And there's no binding in clapack.h.

---John


On Tue, Aug 23, 2011 at 2:27 AM, Sisyphus <address@hidden> wrote:

>
> ----- Original Message ----- From: "John Chludzinski"
>
>  For LAPACK use GotoBLAS2 (http://www.tacc.utexas.edu/**
>> tacc-projects/gotoblas2<http://www.tacc.utexas.edu/tacc-projects/gotoblas2>
>> ).
>> Builds easily with MinGW.
>>
>
> I'm a little bit interested in being able to build the perl extension
> PDL::LinearAlgebra on Windows, but I haven't yet obtained the requisite
> LAPACK library.
> When I try building PDL::LinearAlgebra against the GotoBLAS2 library I get:
>
> Trans.o:Trans.c:(.text+0xd9b): undefined reference to `dlange_'
> Trans.o:Trans.c:(.text+0x29cc)**: undefined reference to `zlange_'
>
> Trans.c declares the following:
>
> /********************/
> extern double dlange_(char *norm, integer *m, integer *n, double *a,
> integer
>  *lda, double *work);
>
> extern double zlange_(char *norm, integer *m, integer *n, dcomplex *a,
>  integer *lda, double *work);
> /********************/
>
> I deduce that the expectation is that the LAPACK library will resolve those
> 2 symbols, but the GotoBLAS2 library clearly doesn't.
>
> Is this a deficiency in the GotoBLAS2 implementation ?
> Or should those symbols be resolved elsewhere ?
>
> (The PDL::LinearAlgebra source is about 4 years old, so I guess there's the
> possibility that this problem arises from a deficiency in *its* source.)
>
> Cheers,
> Rob
>


reply via email to

[Prev in Thread] Current Thread [Next in Thread]