Pierre,
The problem seems to come from the fact that we do not detect how to
generate the assembly code for our atomic operations. As a result we fall
back on the gcc mode for 32 bits architectures.
Here is the corresponding output from the configure script:
checking if cc supports GCC inline assembly... no
checking if cc supports DEC inline assembly... no
checking if cc supports XLC inline assembly... no
checking if CC supports GCC inline assembly... no
checking if CC supports DEC inline assembly... no
checking if CC supports XLC inline assembly... no
checking for assembly format... default-.text-.globl-:--.L-@-1-0-1-1
checking for asssembly architecture... IA32
But there is something strange with the system you're compiling on. The
size of a C pointer is detected as 8 bytes, so this system is running in
64 bits mode. However, configure report it as a i386-pc-solaris2.10
(basically a 32 bits mode).
Give us one or two days to find a fix.
Thanks,
george.
On Mon, 5 Dec 2005, Pierre Valiron wrote:
> Hi,
>
> I tried to experiment open-mpi on our Solaris10 v40z cluster hoping it
> might surpass lam/mpi 7.1.2b28...
>
> I used the following script to compile in 64 bit mode:
>
> #! /bin/tcsh -v
>
> setenv CC "cc"
> setenv CXX "CC"
> setenv FC "f95"
>
> setenv CFLAGS "-O -xtarget=opteron -xarch=amd64"
> setenv CXXFLAGS "-O -xtarget=opteron -xarch=amd64"
> setenv FFLAGS "-O -xtarget=opteron -xarch=amd64"
>
> setenv LDFLAGS "-xtarget=opteron -xarch=amd64"
> setenv CXXLDFLAGS "-xtarget=opteron -xarch=amd64"
>
> ./configure --prefix=/users/valiron/lib/openmpi-1.0.1rc5
> make
>
> The configure runs fine and the make aborts very rapidly.
> I attach the log for configure+make and the config.log file.
>
> Any idea ?
> Pierre.
>
>
"We must accept finite disappointment, but we must never lose infinite
hope."
Martin Luther King
