[petsc-dev] ML used from MATLAB interface

[Barry Smith]

  Dave,

   This is going to be a bitch to fix. The problem is that ML is passing "int" arguments to BLAS and LAPACK (which probably seemed reasonable at the time :-). But the BLAS/LAPACK in MATLAB 64bit is all 64 bit integers. So possible fixes

1) go through all of ML and change all int that are passed to Blas/LAPACK to allow switching to 64 bit int. This ain't going to happen, no one will wade through that code and make these changes.

2) somehow force the BLAS/LAPACK calls from the PETSc shared library to go to the regular BLAS/LAPACK and not Matlab's. I have no idea how to do this :-(.

Note that if hypre/superlu etc use an BLAS/LAPACK calls they __could__ have the same problem (as could the BLAS/LAPACK calls in PETSc) but it seems that sometimes (most of the time?) it uses the "regular" BLAS/LAPACK instead of MATLAB's funny one. I don't understand how or why this happens, perhaps if I understood we could get it to be "always".  It seems to get to the MATLAB ones for "less common" calls.

Note also that PETSc handles the int argument to BLAS/LAPACK with PetscBLASInt which can, in theory be switched to use 64 bit integers and thus happily use the MATLAB BLAS/LAPACK I've tried changing this but have not been happy yet.

To make things even harder MATLAB has three libraries with the word blas in them and 2 with the word lapack.  In addition it appears to use some dynamic loading to get at the mkl libraries. 

bsmith-laptop:petsc-dev barrysmith$ ls /Applications/MATLAB_R2010b.app/bin/maci64/*blas*
/Applications/MATLAB_R2010b.app/bin/maci64/blas.spec
/Applications/MATLAB_R2010b.app/bin/maci64/libmwblas.dylib
/Applications/MATLAB_R2010b.app/bin/maci64/libmwblascompat32.dylib
/Applications/MATLAB_R2010b.app/bin/maci64/refblas.dylib
bsmith-laptop:petsc-dev barrysmith$ ls /Applications/MATLAB_R2010b.app/bin/maci64/*lapack*
/Applications/MATLAB_R2010b.app/bin/maci64/libmwlapack.dylib
/Applications/MATLAB_R2010b.app/bin/maci64/mllapack.dylib

Note also if you think the Mac is bad, it seems to switch to the MATLAB Blas/lapack more under LINUX making the MATLAB interface for PETSc not all that usable at this point, while it is pretty usable from the Apple.

I'll screw around a little more to see if I can understand 2) better. 

  Barry



On Dec 27, 2010, at 11:44 AM, Dave May wrote:

> I'm trying to use PCML via the matlab interface with petsc-dev.
> 
> Every time -pc_type is set to "ml", I find matlab crashes. I've
> attached the matlab script I was using.
> I don't know the "best" way to track this crash down, but I've pasted
> what appears to be the interesting part reported by system stack trace
> (I've attached the entire trace in the rtf file).
> 
> Thread 3:
> 0   libSystem.B.dylib             	0x00007fff84201541 log + 145
> 1   mllapack.dylib                	0x000000014ea229b0 dsteqr_ + 2208
> 2   libpetsc.dylib                	0x000000014e104d3f
> ML_CG_ComputeEigenvalues + 2799 (ml_cg.c:384)
> 3   libpetsc.dylib                	0x000000014e106940 ML_Krylov_Solve
> + 112 (ml_krylov.c:370)
> 4   libpetsc.dylib                	0x000000014e0b0685
> ML_AGG_Gen_Prolongator + 3141 (ml_agg_genP.c:531)
> 5   libpetsc.dylib                	0x000000014e0aeb18
> ML_Gen_MultiLevelHierarchy + 2600 (ml_agg_genP.c:2559)
> 6   libpetsc.dylib                	0x000000014e0af4c4
> ML_Gen_MultiLevelHierarchy_UsingAggregation + 564 (ml_agg_genP.c:2444)
> 7   libpetsc.dylib                	0x000000014da60c91 PCSetUp_ML +
> 3173 (ml.c:601)
> 8   libpetsc.dylib                	0x000000014dc08eeb PCSetUp + 2076
> (precon.c:785)
> 
> It appears that the symbol dsteqr called by ML_CG_ComputeEigenvalues()
> is taken from matlab's own lapack library.
> Does this sound like the right thing which should happen, or is this a
> configuration problem on my end?
> I've attached the configure.log in case.
> 
> I should also point out that I can reliably call hypre through the
> matlab interface without any crashes.
> 
> If there are any tips for debugging this type of error, please let me know.  :)
> And let me know if this type of discussion should be moved over to
> petsc-dev rather than maint.
> 
> Cheers,
>  Dave
> 
> Additional info:
> 
> Matlab version: 7.11.0.584 (R2010b), 64bit
> OS Version: Darwin geop-217.ethz.ch 10.5.0 Darwin Kernel Version
> 10.5.0: Fri Nov  5 23:20:39 PDT 2010;
> root:xnu-1504.9.17~1/RELEASE_I386 i386
> 
> <PCSetUp_ML.rtf><exml.m><configure.log.tar.gz>