[Pw_forum] Re: ask your help!

Tue Oct 26 17:00:22 CEST 2004

Dear Yingwen and Aijun

I've been making some tests on an opteron machines with the CP code contained
in
the Espresso package (last release) and found your make.sys very useful to
improve the one I had. Thank you.

I also found out that with pgi compiler, instead of using acml it is more
convenient to use pgi lapack and goto blas which you can download for free from
the web. I have attached my make.sys to this message.

It's only a small modification to yours (see "LIBS =" part) but the code could
run much faster
(20-25%).

I hope this can help.

Matteo

Quoting Yingwen Song <song at cello.t.u-tokyo.ac.jp>:

> Hi, Aijun
> 
> I am Yingwen Song @ The University of Tokyo. The following are the
> contents of make.sys. Remember to change IFLAGS and LIBS as I told you
> before.
> 
> Wish you good luck.
> 
> 
> Yingwen
> 
> --------------------make.sys START------------------------------
> # make.sys.  Generated from make.sys.in by configure.
> CC             = pgcc
> CCFLAGS        = -O3 -tp k8-64 -fast -Munroll -Mvect=sse
> -Mvect=assoc,cachesize:1048576 -Mcache_align -Kieee -Mvect=prefetch $(DFLAGS)
> $(IFLAGS)
> # See include/defs.h.README for a list of precompilation options
> # (possible arguments to -D or -U) and their meaning
> DFLAGS         = -D__LINUX64 -D__PGI -D__PARA -D__MPI -D__FFTW
> -D__USE_INTERNAL_FFTW
> IFLAGS         = -I. -I../include -I../Modules -I../PW -I../PH
> -I/opt/mpich-1.2.5/include
> CPP            = cpp
> CPPFLAGS       = -P -traditional $(DFLAGS) $(IFLAGS)
> F77            = pgf77
> F90            = pgf90
> FFLAGS         = -fast -r8 -O3 -tp k8-64 -Munroll -Mvect=sse
> -Mvect=assoc,cachesize:1048576 -Mcache_align -Kieee -Mvect=prefetch
> F77FLAGS       = $(FFLAGS) $(IFLAGS)
> F90FLAGS       = $(FFLAGS) $(DFLAGS) $(IFLAGS)
> F77FLAGS_NOOPT = -O0
> LD             = pgf90
> LDFLAGS        =  $(LIBOBJS) $(LIBS)
> LIBOBJS        = ../flib/ptools.a ../flib/flib.a ../clib/clib.a
> # LIBS must contain the location of all needed external libraries
> LIBS           =  -lacml -L/opt/mpich-1.2.5/lib -lfmpich -lmpich
> #LIBS           =  -llapack -lblas
> # MYLIB can be one of the following (depending on LIBS):
> # blas       : compile the local copy of blas routines
> # lapack     : compile the local copy of lapack routines
> # blas_and_lapack : all of the above - use this for a quick test
> #              or if you don't have an optimized blas/lapack library
> # lapack_ibm : compile only lapack routines not present in IBM ESSL
> #              use this together with IBM ESSL
> # lapack_t3e : compile only lapack routines not present in T3E scilib
> #              use this together with T3E scilib
> # lapack_mkl : compile only lapack routines not present in Intel MKL
> #              use this together with Intel MKL
> MYLIB          =
> AR             = ar
> ARFLAGS        = ruv
> --------------------make.sys END------------------------------
> 
> > Hello Yingwen,
> > 
> > Thanks for your timely reply.
> > 
> > I am sorry I couldn't receive the attachment successfully. If possible
> could 
> > you please send to me as mail content or alternative to
> aijun_du at hotmail.com 
> > ?
> > 
> > Many thanks,
> > Aijun
> > ----- Original Message ----- 
> > From: "Yingwen Song" <song at cello.t.u-tokyo.ac.jp>
> > To: "Aijun Du" <a.du at uq.edu.au>
> > Cc: <pw_forum at pwscf.org>
> > Sent: Friday, October 22, 2004 5:11 PM
> > Subject: Re: ask your help!
> > 
> > 
> > > WARNING: This e-mail has been altered by mail systems at
> > > Information Technology Services at The University of Queensland
> > > (www.its.uq.edu.au)
> > > Following this paragraph are indications of the actual
> > > changes made.
> > >
> > > A file named
> > >
> > >  make.sys
> > >
> > > was removed from this email as it constituted a security hazard.
> > >
> > > It may be a virus. If you require this file, please contact
> > > the sender and arrange an alternate means of receiving it.
> > > For example ask the sender to resend it within
> > > a non self-extracting archive.
> > >
> > > More information can be found at:
> > >
> > > http://www.uq.edu.au/uqnet/virus.html
> > >
> > >
> > 
> > 
> >
>
--------------------------------------------------------------------------------
> > 
> > 
> > > Hi, Aijun
> > >
> > > I am Yingwen Song @ The University of Tokyo. The MPI version of make.sys
> > > is attached. I am using MPICH 1.2.5. Please modify
> > > "-I/opt/mpich-1.2.5/include"
> > > in IFLAGS and
> > > "-L/opt/mpich-1.2.5/lib"
> > > in LIBS
> > >
> > > to point to your own MPI path.
> > >
> > > Good luck.
> > >
> > >
> > >> Dear Dr. Song,
> > >>
> > >> I am working in the University of Queensland and also using pwscf for my
> 
> > >> research work. I find your email address from PWSCF email list. I knew 
> > >> you have succeeded in compiling and running PWSCF on AMD opetron64 using
> 
> > >> pgf90. Have you tried MPI version? Could you send me the Make.sys(MPI)
> to 
> > >> me?
> > >>
> > >> I have compiled pwscf on my new AMD opteron64 machine. My test on a big
> 
> > >> system met some problems and stopped.  But for small system it is okay.
> > >>
> > >> Many thanks,
> > >> Regards,
> > >> Aijun
> > >
> > > --------------------------
> > > Yingwen Song, Ph.D.
> > > The University of Tokyo
> > > Email:song at cello.t.u-tokyo.ac.jp
> > > Homepage:http://cello.t.u-tokyo.ac.jp/~song/
> > > Tel/Fax: 03-5841-1286
> > >
> > > 
> 
> --------------------------
> Yingwen Song, Ph.D.
> The University of Tokyo
> Email:song at cello.t.u-tokyo.ac.jp
> Homepage:http://cello.t.u-tokyo.ac.jp/~song/
> Tel/Fax: 03-5841-1286
> 
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: make.sys_opt_pgi_ser_goto_lap
Type: application/octet-stream
Size: 2125 bytes
Desc: not available
Url : /pipermail/attachments/20041026/cb8922ba/attachment.obj 