This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Irix6 long doubles implemented wrong? (27_io/ostream_inserter_arith)


On Jan 28, 2003, "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu> wrote:

> PS: over here:
> http://gcc.gnu.org/ml/gcc/2002-12/msg01310.html
> you mentioned something about using the native irix library for some
> of the math to speed it up.  What needs to be done there?  Can you
> explain it to me (somewhat verbosely? :-)

Have a look at the macro GOFAST_RENAME_LIBCALLS in
gcc/config/gofast.h.  gofast is a FP emulation library that uses entry
points different from those used by GCC, so GOFAST_RENAME_LIBCALLS
arranges for GCC to issue calls using these entry points instead of
those it traditionally does.

You assignment :-) is to: (i) find out the name of the functions that
IRIX uses for long double floating-point operations and (ii) add
similar assignments to the optab arrays and libfunc variables such
that we issue calls to the IRIX-specific names.  An optional (iii) is
to change the name of the functions emitted for the emulation library
in gcc/config/fp-bit.h in the case of __LDBL_MANT_DIG__ == 106, and/or
disable it completely on IRIX (removing tp-bit.c from t-iris6), so
that we use only the native implementation.

Please let me know if you could use some further clarification of any
of the items above, and thanks in advance if you have a chance of
looking into them.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]