Lapack install

Open discussion regarding features, bugs, issues, vendors, etc.

Lapack install

Postby moulaye » Tue Mar 21, 2006 12:33 pm

Hy,

I try to install Lapack on my computer ( hp pavillon zv5000, OS : Linux fedora core (2.6.9.- 1.667) ). After typing "make all" all source are compiled but the execution of testing executables is not done (ex : xlintsts)
Is there somebody who can help me ?
Excuse for my english
Thanks.
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Tue Mar 21, 2006 3:17 pm

Hello,
[ Thanks for the effort to speak English !]

that's weird.
What message do you have?
Are you speaking of the testing or the timing?

If you want to force the testing, you can go in the TESTING directory and type 'make' (that should have been done automatically however).

Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Lapack install

Postby moulaye » Tue Mar 21, 2006 5:58 pm

Hello,
I speak about testing .
In the directory TESTING there is only two executables: xlintsts and xeigtsts. After type make the execution of one executable (xlintsts or xeigtsts) begin, but after several minutes not of result. What appears abnormal to me I force the program to stop.

Thanks for your answer.
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Tue Mar 21, 2006 6:11 pm

so there should have eight executables not two

xlintsts (single - linear system)
xlintstd (double - linear system)
xlintstc (single complex - linear system)
xlintstz (double complex - linear system)
xeigtsts (single - eigen system)
xeigtstd (double - eigen system)
xeigtstc (single complex - eigen system)
xeigtstz (double complex - eigen system)

so first you are missing some buddies....


then try
cd TESTING
./xlintsts < stest.in

this works?

if yes try:
./xeigtsts < nep.in
./xeigtsts < sep.in
./xeigtsts < svd.in
./xeigtsts < sec.in
./xeigtsts < sed.in
./xeigtsts < sgg.in
./xeigtsts < sgd.in
./xeigtsts < ssb.in
./xeigtsts < ssg.in
./xeigtsts < sbal.in
./xeigtsts < sbak.in
./xeigtsts < sgbal.in
./xeigtsts < sgbak.in
./xeigtsts < sbb.in
./xeigtsts < glm.in
./xeigtsts < gqr.in
./xeigtsts < gsv.in
./xeigtsts < lse.in

if this works, I think I have a guess...
what's your shell?

echo $SHELL

??

Anyway just try:

setenv SHELL bash; make

we are using redirection in the command line that are not standard for some SHELL (csh for example)

Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Lapack install

Postby moulaye » Tue Mar 21, 2006 7:12 pm

Hello Julien,

I followed your indiactions but i have the same problem.
After typing the command setenv SHELL bash;make
First i have this message: "bash: setenv: command not found",
The compilation process continue, the execution begin by :
./xeigtsts < sdv.in > ssvd.out 2>&1
and no result appears.
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Tue Mar 21, 2006 10:00 pm

no need to set your SHELL to bash, you are already using bash, so forget about the setenv.

ok so you should have a bunch of *.out files that have been created no?

If the makefile is at the step you describe:
./xeigtsts < sdv.in > ssvd.out 2>&1
this means that you should have
stest.out
snep.out
ssep.out
for sure in the TESTING directory.

Open those files, everything looks ok?

then if the make does not work... what does all this give (see below)?
I hope this should be ok.
If something hangs let me know which one and please try the others.

./xeigtsts < nep.in
./xeigtsts < sep.in
./xeigtsts < svd.in
./xeigtsts < sec.in
./xeigtsts < sed.in
./xeigtsts < sgg.in
./xeigtsts < sgd.in
./xeigtsts < ssb.in
./xeigtsts < ssg.in
./xeigtsts < sbal.in
./xeigtsts < sbak.in
./xeigtsts < sgbal.in
./xeigtsts < sgbak.in
./xeigtsts < sbb.in
./xeigtsts < glm.in
./xeigtsts < gqr.in
./xeigtsts < gsv.in
./xeigtsts < lse.in
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

lapck install

Postby moulaye » Wed Mar 22, 2006 7:28 am

Hello,
There are many error messages in those 3 files (i don't understand those message).
For example in stest.out all test value are INF for SGE routine,is it normal?
At beigin of the file (snep.out,ssep.out,stest.out) there are (i thing ) machine' s parameters:
Relative machine underflow is taken to be 0.000000E+00
Relative machine overflow is taken to be -0.368935E+20
Relative machine precision is taken to be 0.000000E+00
Those parameters are normal?
Because i don't know how to attach a file to this message ,i can't send you those files(snep.out,ssep.out,stest.out).
Thanks again for help
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Wed Mar 22, 2006 9:37 am

hello,
so yes you can stop everything. this is not good at all.
Can you just copy-paste your make.inc please?
Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Postby moulaye » Wed Mar 22, 2006 10:20 am

Hello,
This is a file make.inc
####################################################################
# LAPACK make include file. #
# LAPACK, Version 3.0 #
# June 30, 1999 #
####################################################################
#
SHELL = /bin/sh
#
# The machine (platform) identifier to append to the library names
#
PLAT = _LINUX
#
# Modify the FORTRAN and OPTS definitions to refer to the
# compiler and desired compiler options for your machine. NOOPT
# refers to the compiler options desired when NO OPTIMIZATION is
# selected. Define LOADER and LOADOPTS to refer to the loader and
# desired load options for your machine.
#
FORTRAN = g77
OPTS = -funroll-all-loops -fno-f2c -O3
DRVOPTS = $(OPTS)
NOOPT =
LOADER = g77
LOADOPTS =
#
# The archiver and the flag(s) to use when building archive (library)
# If you system has no ranlib, set RANLIB = echo.
#
ARCH = ar
ARCHFLAGS= cr
RANLIB = ranlib
#
# The location of the libraries to which you will link. (The
# machine-specific, optimized BLAS library should be used whenever
# possible.)
#
BLASLIB = ../../blas$(PLAT).a
LAPACKLIB = lapack$(PLAT).a
TMGLIB = tmglib$(PLAT).a
EIGSRCLIB = eigsrc$(PLAT).a
LINSRCLIB = linsrc$(PLAT).a
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Wed Mar 22, 2006 10:52 am

I really do not know what's the problem.
The installation is pretty standard on all the Linux boxes and your make.inc is ok.

(1) so you can do a make clean of everything, nothing's good here.
(2) can you give me the output of

make install

(3) what is your machine
(4) what is your compiler version

Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

lapck install

Postby moulaye » Wed Mar 22, 2006 1:44 pm

Here is the out put file of make install
( cd INSTALL; make; ./testlsame; ./testslamch; \
./testdlamch; ./testsecond; ./testdsecnd; \
cp lsame.f ../BLAS/SRC/; cp lsame.f ../SRC; \
cp slamch.f ../SRC/; cp dlamch.f ../SRC/; \
cp second.f ../SRC/; cp dsecnd.f ../SRC/ )
make[1]: Entering directory `/home/mahamane/LAPACK/INSTALL'
g77 -funroll-all-loops -fno-f2c -O3 -c lsame.f
g77 -funroll-all-loops -fno-f2c -O3 -c lsametst.f
g77 -o testlsame lsame.o lsametst.o
g77 -c slamch.f
g77 -funroll-all-loops -fno-f2c -O3 -c slamchtst.f
g77 -o testslamch slamch.o lsame.o slamchtst.o
g77 -c dlamch.f
g77 -funroll-all-loops -fno-f2c -O3 -c dlamchtst.f
g77 -o testdlamch dlamch.o lsame.o dlamchtst.o
g77 -funroll-all-loops -fno-f2c -O3 -c second.f
g77 -funroll-all-loops -fno-f2c -O3 -c secondtst.f
g77 -o testsecond second.o secondtst.o
g77 -funroll-all-loops -fno-f2c -O3 -c dsecnd.f
g77 -funroll-all-loops -fno-f2c -O3 -c dsecndtst.f
g77 -o testdsecnd dsecnd.o dsecndtst.o
g77 -funroll-all-loops -fno-f2c -O3 -c tstiee.f
g77 -o testieee tstiee.o
make[1]: Leaving directory `/home/mahamane/LAPACK/INSTALL'
ASCII character set
Tests completed
Epsilon = 0.
Safe minimum = 0.
Base = 0.
Precision = 0.
Number of digits in mantissa = 0.
Rounding mode = 0.
Minimum exponent = 0.
Underflow threshold = 0.
Largest exponent = 0.
Overflow threshold = -3.68934881E+19
Reciprocal of safe minimum = INF
Epsilon = 1.11022302E-16
Safe minimum = 2.22507386E-308
Base = 2.
Precision = 2.22044605E-16
Number of digits in mantissa = 53.
Rounding mode = 1.
Minimum exponent = -1021.
Underflow threshold = 2.22507386E-308
Largest exponent = 1024.
Overflow threshold = 1.79769313E+308
Reciprocal of safe minimum = 4.49423284E+307
Time for 1,000,000 SAXPY ops = 0.200E-02 seconds
SAXPY performance rate = 500. mflops
Including SECOND, time = 0.200E-02 seconds
Average time for SECOND = 0.200E-06 milliseconds
Equivalent floating point ops = 0.100 ops
Time for 1,000,000 DAXPY ops = 0.300E-02 seconds
DAXPY performance rate = 333. mflops
Including DSECND, time = 0.500E-02 seconds
Average time for DSECND = 0.400E-03 milliseconds
Equivalent floating point ops = 133. ops
My machine is hp pavillon zv5000 ,the procesor is AMD 64 Athlon
The OS is Linux fedora core (2.6.9 -1.667) ,i don't know what version of compilator is associated to this lLinux version.
Thank you very much.
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Wed Mar 22, 2006 1:58 pm

pefect... so you definetely have a problem,
it's important for us to know the version of the compiler, just type
g77 --version
It's nice if you can answer. It will help us.

So if you are running out of time with this installation, what you can do
id just install the RPM. I am sure Fedora provides good RPM for lapack and
reference blas. (It's a bit the solution of the poor though, sorry about that.)

Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Postby moulaye » Wed Mar 22, 2006 3:22 pm

Typing command g77 --version gave this result
GNU Fortran (GCC) 3.4.2 20041017 (Red Hat 3.4.2-6.fc3)
Copyright © 2004 Free Software Foundation, Inc.

What is RPM ?
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby moulaye » Tue Mar 28, 2006 4:31 am

Hello,
Is in lapack a routine which solve a linear system with iterative method (ex: bicgstab or gmres)? :?:
moulaye
 
Posts: 9
Joined: Mon Mar 20, 2006 10:37 am
Location: paris

Postby Julien Langou » Tue Mar 28, 2006 9:09 am

No there is not. LAPACK is focused on dense linear algebra problems (i.e. problems where all the elements of the matrix are stored in some predefined storage format GE, BD, SY, SP,....).
Iterative methods (GMRES, BiCGStab) are less attractive for this class of problems. You will find a software review on the web page:
http://www.netlib.org/utk/people/JackDongarra/la-sw.html
if you go in the item 'sparse iterative solvers' and just make your pick.
Julien
Julien Langou
 
Posts: 832
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Next

Return to User Discussion

Who is online

Users browsing this forum: Google [Bot] and 2 guests