This is the mail archive of the gcc@gcc.gnu.org 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]

gcc-3.1 bootstrap (all languages) successful with -O3 -funroll-loops



gcc-3.1 bootstrap (all languages) successful with -O3 -funroll-loops

# output of config.guess:
sparc-sun-solaris2.6

# output gcc -v:
Reading specs from /usr/local/links/bin/../lib/gcc-lib/sparc-sun-solaris2.6/3.1/specs
Configured with: ../gcc-3.1-static-bimbo2.6/configure --prefix=/usr/local/gcc/gcc-3.1 --disable-shared --with-gnu-as --with-as=/usr/local/bin/as --with-gnu-ld --with-ld=/usr/local/bin/ld --enable-threads --with-cpu=supersparc --enable-cpp --enable-version-specific-runtime-libs --disable-nls
Thread model: posix
gcc version 3.1

# languages
all (full distribution)

# sytem info
Solaris 2.6 (with recent patches) on Sun Ultra-1;
the produced binary installation is also successfully used
on Sparcstation-4 (Solaris 2.6), which is why -msupersparc was used,
and also to be used on Ultra-10 (Solaris 2.7/8)

# other info
in contrast to earlier versions (including gcc-3.0.4), everything
and all languages built fine even using non-standard options
for the bootstrap:
    -fPIC -O3 -funroll-loops -msupersparc -Qn -Wa,-Qn -s
However, the testsuite results (submitted using the script)
deviate slightly from the ones reported for Solaris2.6
(presumably bootstrapped with more standard options).
***** Why do the compiler flags used for the bootstrap *****
***** do not get reported with gcc -v, and are also    *****
***** not required for reporting?                      *****
(it seems to make a difference, even regarding to the success
of the bootstrap, resp. of the stage2/3 comparison, which failed
with gcc-3.0.4 with the above flags...)

Arto Nirkko


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