| Line 5118... |
Line 5118... |
| 5118 |
number) as well as specific details of rounding. Although almost all
|
5118 |
number) as well as specific details of rounding. Although almost all
|
| 5119 |
current FPUs can support this, selecting such support can be a pain.
|
5119 |
current FPUs can support this, selecting such support can be a pain.
|
| 5120 |
The problem is that there is no agreement on how to set the signalling
|
5120 |
The problem is that there is no agreement on how to set the signalling
|
| 5121 |
behaviour; Sun/Sparc, SGI/IRIX and @cputype{ix86} Linux require no
|
5121 |
behaviour; Sun/Sparc, SGI/IRIX and @cputype{ix86} Linux require no
|
| 5122 |
special action, FreeBSD requires a call to (the macro)
|
5122 |
special action, FreeBSD requires a call to (the macro)
|
| 5123 |
@code{fpsetmask(0)} and OSF1 requires that computation be done with a
|
5123 |
@code{fpsetmask(0)} and OSF1 required that computation be done with a
|
| 5124 |
@option{-ieee_with_inexact} flag etc. On a new platform you must find
|
5124 |
@option{-ieee_with_inexact} flag etc. On a new platform you must find
|
| 5125 |
out the magic recipe and add some code to make it work. This can often
|
5125 |
out the magic recipe and add some code to make it work. This can often
|
| 5126 |
be done via the file @file{config.site} which resides in the top level
|
5126 |
be done via the file @file{config.site} which resides in the top level
|
| 5127 |
directory.
|
5127 |
directory.
|
| 5128 |
|
5128 |
|
| 5129 |
Beware of using high levels of optimization, at least initially. On
|
5129 |
Beware of using high levels of optimization, at least initially. On
|
| 5130 |
many compilers these reduce the degree of compliance to the
|
5130 |
many compilers these reduce the degree of compliance to the
|
| 5131 |
@acronym{IEEE} model. For example, using @option{-fast} on the Solaris
|
5131 |
@acronym{IEEE} model. For example, using @option{-fast} on the Solaris
|
| 5132 |
Studio compilers has caused @R{}'s @code{NaN} to be set incorrectly.
|
5132 |
Studio compilers has caused @R{}'s @code{NaN} to be set incorrectly, and
|
| - |
|
5133 |
@command{gcc}'s @option{-ffast-math} has given incorrect results.
|
| 5133 |
|
5134 |
|
| 5134 |
@strong{Shared Objects}: There seems to be very little agreement
|
5135 |
@strong{Shared Objects}: There seems to be very little agreement
|
| 5135 |
across platforms on what needs to be done to build shared objects.
|
5136 |
across platforms on what needs to be done to build shared objects.
|
| 5136 |
there are many different combinations of flags for the compilers and
|
5137 |
there are many different combinations of flags for the compilers and
|
| 5137 |
loaders. @acronym{GNU} libtool cannot be used (yet), as it currently
|
5138 |
loaders. @acronym{GNU} libtool cannot be used (yet), as it currently
|