This is the mail archive of the mailing list for the GCC 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]

target/6565: x86 : -msse does not define __SSE__

>Number:         6565
>Category:       target
>Synopsis:       x86 : -msse does not define __SSE__
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun May 05 10:26:00 PDT 2002
>Originator:     Sylvain Pion
>Release:        3.1 and 3.2
Reading specs from ./gcc/Linux_trunc/bin/../lib/gcc-lib/i686-pc-linux-gnu/3.2/specs
Configured with: /u/zosma/0/prisme/spion/gcc/gcc_trunc/configure --disable-shared --disable-nls --enable-languages=c++ --prefix=/u/zosma/0/prisme/spion/gcc/Linux_trunc --with-gnu-as --with-gnu-ld --with-as=/u/zosma/0/prisme/spion/gcc/Binutils/Linux/bin/as --with-ld=/u/zosma/0/prisme/spion/gcc/Binutils/Linux/bin/ld
Thread model: single
gcc version 3.2 20020502 (experimental)
-D__SSE__ and -D__SSE2__ are only added with -march=pentium4,
but -msse or -msse2 ought to be enough since they are enough to generate SSE code when -mfpmath=sse is set.

On the other hand, -march=pentium4 alone (without -mfpmath=sse) defines __SSE__, but uses the i387 for math.

See the thread starting at :

In short, there should be a way (macro) to tell the programmer if SSE[2] math is going to be used instead of i387 math (which is not IEEE 754 compliant, and thus requires costly workarounds in some cases).
touch foo.c
gcc -v -c -msse -mfpmath=sse foo.c
=> does not define -D__SSE2__



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