|
From: | Michael Baginski |
Subject: | problems installing octave-forge on Altix machine (64 bit) |
Date: | Mon, 11 Jul 2005 16:19:24 -0500 |
Hello Octave help
users-groups, In the last several weeks I have
been working with my associates at the Sincerely, Mike The attached message from Charles
Wright: Michael E. Baginski, Ph.D.
Associate Professor of Electrical and Comp. Engineering Dept. of Electrical and Computer Engineering. Room 423 Broun Hall Auburn University Auburn, Alabama 36849 e-mail: address@hidden phone: 334-844-1800 Mike,
Attached are my notes on 9.2.3. I don't have the build log for the other version. Can you get me compile notes from whoever has this working on the
Altix? I've tried
everything I can think of. If you find someone willing to help I can set them up an account on the altix so they can compile it on our machine. They can take my attached notes and generate the build logs in their home directory. I tried the beta 9.2.3 version the developer recommended, can't seemed to make it happy with umfpack. (probably has to do with the immature beta state) I did get umfpack to compile however octave 9.2.3 gives this. LC_ALL=POSIX export LC_ALL LDFLAGS="-L/opt/asn/lib" \ CFLAGS="-fPIC -I/opt/asn/include" \ FFLAGS="-I/opt/asn/include" \ CPPFLAGS="-I/opt/asn/include" \ CXXFLAGS="-I/opt/asn/include" \ ./configure --enable-dl --enable-shared=yes --enable-rpath --enable-lite-kernel --enable-picky-flags --enable-static=no --with-g77 --prefix=/opt/asn/apps/octave-2.9.3 (I had to create /opt/asn/include/umfpack for it to find the header files but still got this... Maybe a bug in the configure script.) checking umfpack/umfpack.h usability... yes checking umfpack/umfpack.h presence... yes checking for umfpack/umfpack.h... yes checking for amd_postorder in -lamd... yes checking for umfpack_zi_get_determinant in -lumfpack... no checking for umfpack_zi_get_determinant in -lumfpack... no checking for umfpack_zi_get_determinant in -lumfpack... no ./configure: line 13119: ,: command not found Here's the make file for UMFPACK that I used. UMFPACK was painful as well and I'm not sure this is 64 bit... Demos were segfaulting with the 64bit version of scsl address@hidden Make]# cat Make.linux #------------------------------------------------------------------------------ # Linux configuration (for both AMD and UMFPACK) #------------------------------------------------------------------------------ # You may also need to add -lunwind -lcprts to the LIB= string, # if you use the Intel compiler and the Fortran BLAS. # Using GNU gcc and f77 compilers: # CC = gcc # CFLAGS = -O2 -fPIC -DLP64 # F77FLAGS = -64 # Using Intel's icc and ifc compilers: F77 = ifort CC = icc CFLAGS = -ansi -O2 #CFLAGS = -pg -g # Using gcc compiler with picky tests # CC = gcc # CFLAGS = -ansi -pedantic -W -Wall -Wno-parentheses -Wshadow -Wcast-align -Winline -Wstrict-prototypes -Wno-unknown-pragmas -O3 -fPIC # for gcc and gcov: # CC = gcc # CFLAGS = -pg -ftest-coverage -fprofile-arcs # Running splint # CC = - splint # CFLAGS = -weak -fixedformalarray -badflag -fileextensions +relaxtypes +posixlib -unrecog #------------------------------------------------------------------------------ # BLAS options (for UMFPACK only) #------------------------------------------------------------------------------ # 1: with no BLAS (this will be slow) # CONFIG = -DNBLAS # LIB = -lm # 2: with the ATLAS C-BLAS (http://www.netlib.org/atlas). # CONFIG = -DCBLAS -I../ATLAS/include # LIB = -lcblas -latlas -lm # 3: with Fortran interface to the ATLAS BLAS # CONFIG = # LIB = -lf77blas -latlas -lfrtbegin -lg2c -lm # 4: with Fortran interface to the BLAS, and Goto's BLAS # CONFIG = # LIB = -lgoto -lxerbla -lfrtbegin -lg2c -lm # 5: with Fortran interface to the BLAS, and Goto's BLAS, and no divide-by-zero # CONFIG = -DNO_DIVIDE_BY_ZERO # LIB = -lgoto -lxerbla -lfrtbegin -lg2c -lm #6: Altix 350 scsl CONFIG = -DCBLAS LIB = -lscs -lm -- Charles Wright, HPC Systems Administrator Alabama Research and Education Network Computer Sciences Corporation |
octave-compile2.txt
Description: Text document
Michael Baginski.vcf
Description: Text document
[Prev in Thread] | Current Thread | [Next in Thread] |