Accessing VECLIB

For example, the command lines in Method 2 for PA could be written:

f90 [options] file ... −Wl,–aarchive_shared,−L/opt/mlib/lib/[pa2.0pa20_64] −lveclib

cc[options] file ... −Wl,–aarchive_shared,−L/opt/mlib/lib/[pa2.0pa20_64] −lveclib −lcl −lm

aCC [options] file ... −Wl,–aarchive_shared,−L/opt/mlib/lib/[pa2.0pa20_64] −lveclib −lcl −lm

4.Set the LDOPTS environment variable to include:

–aarchive_shared,−L/opt/mlib/lib/[pa2.0pa20_64] For example:

setenv LDOPTS “–aarchive_shared,–L/opt/mlib/lib/pa2.0”

Then use the −lveclib option on the compiler command line that links your program:

f90 [options] file ... −lveclib

cc [options] file ... −lveclib −lcl −lm

aCC [options] file ... −lveclib −lcl −lm

NOTE

An LDOPTS specification takes precedence over using -Wlon the

 

compiler command line. That is, if you use the LDOPTS environment

 

variable to specify a library path, you cannot override that specification

 

with a -Wloption on your compiler command line.

 

 

5.To link with the VECLIB8 libraries on a PA-based system, use the +DA2.0W option to specify memory addresses are 64-bit.

Compile Fortran applications with the +i8, +autodbl, or +autodbl4 option where:

•+i8 promotes 4-byte integer of logical constants, intrinsics, and user variables (declared as integer or logical) to 8-byte quantities.

•+autodbl promotes all integer, logical, and real items to 8 bytes, and all double-precision and complex items to 16 bytes.

•+autodbl4 promotes all integer logical, and real items to 8 bytes, and complex items to 16 bytes. The +autodbl4 option does not promote the size of double-precision and double-complex items.

f90 +DA2.0W +i8 [options] file ... −Wl,–aarchive_shared −lveclib8

cc+DA2.0W [options] file ... −Wl,–aarchive_shared −lveclib8 −lcl −lm aCC +DA2.0W [options] file ... −Wl,–aarchive_shared −lveclib8 −lcl −lm

Chapter 1 Introduction to VECLIB 7