[Wien] undefined function calls when linking to BLAS/LAPACK

[kjharris]

The linker options you have below point at both the ATLAS libs you compiled 
and at the slow, unoptimized liblapack_lapw.a located in <WIEN_ROOT>/SRC_lib. 
Could you check and see if it compiles w/o this to see if it works with a 
compiled ATLAS version.

also, I believe atlas defaults to -nofor_main when it sees you trying to 
compile with the intel fortran compiler

I think you may be taking the right approach though, I only read the first 
part of the intel-gcc compatibility white paper but after your email, I looked 
farther on and realized that binary compatibility only applies to C code , not 
Fortran.  I had originally compiled ATLAS with icc/g77 because the webpage 
notes that only interface routines are actually compiled in fortran. I am 
currently attempting to compile ATLAS with the intel fortan compiler to see if 
this gets me anywhere.

also, I just noticed a note on the ATLAS site that says order of libraries 
linked in is important (this does not solve my problem however)

p.s., I can't figure out what the -Vaxlib option is for, anyone know?



>There may be further problems down the road, but at least Wien2k compiled
>without errors at the end of ./siteconf_lapw. The otions settings for Wien2k 
were:
>
>current:FOPT:-FR -mp -w -nofor_main
>current:FPOPT:-FR -mp -w
>current:LDFLAGS:-L../SRC_lib -L/usr/lib -L/opt/ATLAS/lib/lin-K7 -Vaxlib 
-static
>current:DPARALLEL:'-DParallel'
>current:R_LIBS:-llapack_lapw -lf77blas -lcblas -latlas -lguide -lpthread
>
>I recall that there was some mail by Lutz Rausch that -static leeds to errors
>with intels mkl 6.1, and I did not test this built in any way up to now (just
>compiled, sorry), so there may be plenty of problems ahead.
>
>Please inform me, when (and how) you succeeded, just in case I get stuck on 
the
>way again (I also have to get this thing running soon).
>
>Martin
>