[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Installing Octave Interface for Liblinear
From: |
Freddy López |
Subject: |
Re: Installing Octave Interface for Liblinear |
Date: |
Wed, 9 Mar 2011 00:49:33 -0400 |
Hello,
And, how to compile LIBLINEAR under Windows?
Although this is not an exactly answer to sazerac, it will be useful
for some people are looking for it by there.
I modified the make.m file to this one:
% This make.m is used under Windows
mex -c '..\blas\daxpy.c'
mex -c '..\blas\ddot.c'
mex -c '..\blas\dnrm2.c'
mex -c '..\blas\dscal.c'
mex -c '..\linear.cpp'
mex -c '..\tron.cpp'
mex "-Wp,-DMX_API_VER=0x07030001" -c '..\linear.cpp'
mex "-Wp,-DMX_API_VER=0x07030001" -c '..\tron.cpp'
mex "-Wp,-DMX_API_VER=0x07030001" -c linear_model_matlab.c -I..\\
mex "-Wp,-DMX_API_VER=0x07030001" train.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\dscal.o'
mex "-Wp,-DMX_API_VER=0x07030001" train.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\dnrm2.o'
mex "-Wp,-DMX_API_VER=0x07030001" train.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\ddot.o'
mex "-Wp,-DMX_API_VER=0x07030001" train.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\daxpy.o'
mex "-Wp,-DMX_API_VER=0x07030001" predict.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\dscal.o'
mex "-Wp,-DMX_API_VER=0x07030001" predict.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\dnrm2.o'
mex "-Wp,-DMX_API_VER=0x07030001" predict.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\ddot.o'
mex "-Wp,-DMX_API_VER=0x07030001" predict.c -I..\\ '..\tron.o'
'..\linear.o' linear_model_matlab.o '..\blas\daxpy.o'
mex "-Wp,-DMX_API_VER=0x07030001" libsvmread.c
mex "-Wp,-DMX_API_VER=0x07030001" libsvmwrite.c
and it run perfectly (tested with the README file's example inside
liblinear folder).
Cheers.
On Wed, Feb 23, 2011 at 08:25, Andy Buckle <address@hidden> wrote:
>
> On Wed, Feb 23, 2011 at 12:00 PM, Benjamin Lindner <address@hidden> wrote:
> >> Thank you for your response. I think "-O" stands for "optimize".
> >
> > Oh, I see, it's a gcc option. Well these are not supported in mkoctfile.
> > Neither, by the way, is the flag -largeArrayDims (it's interpreted as
> > adding the library libargeArrayDims to the linker command line), the
> > flag -outdir, and wildcard patterns for the input.
> > So you can't use this make.m file with octave - you have to rewrite it
> > to use one statement per file, and remove the unsupported flags, and
> > remove the double-definition error on "mwIndex", and either escape the
> > backslash or use single quotes, like such
> >
> > mex -c '..\blas\daxpy.c'
> > mex -c '..\blas\ddot.c'
> > mex -c '..\blas\dnrm2.c'
> > mex -c '..\blas\dscal.c'
> > mex -c '..\linear.cpp'
> > mex -c '..\tron.cpp'
> > mex -DMX_API_VER=0x07030000 -c linear_model_matlab.c -I..\\
> > mex -DMX_API_VER=0x07030000 train.c -I..\\ '..\tron.o' '..\linear.o'
> > linear_model_matlab.o '..\blas\daxpy.o' '..\blas\ddot.o'
> > '..\blas\dnrm2.o' '..\blas\dscal.o'
> > mex -DMX_API_VER=0x07030000 predict.c -I..\\ '..\tron.o' '..\linear.o'
> > linear_model_matlab.o '..\blas\daxpy.o' '..\blas\ddot.o'
> > '..\blas\dnrm2.o' '..\blas\dscal.o'
> > mex -DMX_API_VER=0x07030000 libsvmread.c
> > mex -DMX_API_VER=0x07030000 libsvmwrite.c
> >
> > Now compilation works for me.
> > Mind, I can't test whether the functions perform now as expected,
> > being no expert on the topic, so you have to do this yourself!
> >
> > benjamin
>
> When call mkoctfile with -v, I see the underlying compiler call is
> using -O3. (sf.net 3.2.4 installer on windows xp).
>
> >mkoctfile -v vec_fun_practice.cpp
> mingw32-g++-4.4.0-dw2 -c -I/usr/local/octmgw32_gcc-4.4.0-dw2/include
> -IC:\Octav
> e\3.2.4_gcc-4.4.0\include\octave-3.2.4
> -IC:\Octave\3.2.4_gcc-4.4.0\include\octav
> e-3.2.4\octave -IC:\Octave\3.2.4_gcc-4.4.0\include -mieee-fp -D_GLIBCXX_DLL
> -ma
> rch=i686 -mtune=generic -O3 -Wall vec_fun_practice.cpp -o
> vec_fun_practice.o
> mingw32-g++-4.4.0-dw2 -shared -Wl,--export-all-symbols
> -Wl,--enable-auto-import
> -Wl,--enable-runtime-pseudo-reloc -shared-libgcc
> -Wl,--allow-multiple-definition
> -o vec_fun_practice.oct vec_fun_practice.o
> -LC:\Octave\3.2.4_gcc-4.4.0\lib\
> octave-3.2.4 -LC:\Octave\3.2.4_gcc-4.4.0\lib -loctinterp -loctave -lcruft
> -shar
> ed-libgcc -Wl,--allow-multiple-definition -llapack -lblas -lfftw3 -lfftw3f
> -lrea
> dline -lncurses -liberty -lhdf5 -lz -lm -lgdi32 -lws2_32 -luser32
> -lkernel32 -
> lgfortran -lstdc++_s
>
> --
> /* andy buckle */
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://mailman.cae.wisc.edu/listinfo/help-octave
--
«But Gwindor answered: 'The doom lies in yourself, not in your name.'»
JRR Tolkien
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: Installing Octave Interface for Liblinear,
Freddy López <=