Catastrophic Error with "-m32" flag

[gle6b]

I have to deploy my code on a grid computing framework. I used to compile the code on my 64-bit machine, but lots of machine on the grid are 32-bit.

When I try to compile the code with the "-m32" flag, I get:

"

ifort -c -m32 orig_flame.f

orig_flame.f(159): (col. 9) remark: LOOP WAS VECTORIZED.

0_1529

: catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error.

in file orig_flame.f, line 2, column 15

compilation aborted for orig_flame.f (code 3)

ifort -c -m32 orig_flame.forig_flame.f(159): (col. 9) remark: LOOP WAS VECTORIZED.0_1529
: catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error.in file orig_flame.f, line 2, column 15
compilation aborted for orig_flame.f (code 3)

"

My machine:

Linux *********** 2.6.18-164.6.1.el5.centos.plus #1 SMP Wed Nov 4 09:31:39 EST 2009 x86_64 x86_64 x86_64 GNU/Linux

Compiler Version:

Intel Fortran Intel 64 Compiler Professional for applications running on Intel 64, Version 11.0 Build 20090131 Package ID: l_cprof_p_11.0.081

I do not have root privileges on the machine.

[Ron_Green]

the ICE or "Internal Compiler Error" is a general error message covering any circumstance where the compiler crashes.

Your compiler is a bit old, can you get a newer one from https://registrationcenter.intel.com and try with a recent 11.1? Or better yet, follow the instructions at the top of this forum and post the file orig_flame.f and any modules or include files that it may reference. I can test this with your compiler version and the most recent versions.

thanks

ron

[gle6b]

ron,

thanks for the prompt reply. I will see if I can ask the system administrator to upgrade the Intel compiler to the newer version. If something keeps on going wrong, I will strip the code by some copyrighted parts and post it on the forum.

Thanks,

g

[TimP]

The Intel compilers don't observe a -m32 option such as gcc does. Even for gcc, that switch turns over control to a separate 32-bit compiler, if you have installed it; it does not cause the 64-bit compiler to work with 32-bit code.

The failure you saw appears to be a bug, but not necessarily directly related to what you are trying to do.

For 32-bit compatible compilation, you must set environment PATH variables in accordance with the ifortvars 32-bit script e.g. 'source /opt/intel/Compiler/11.1/064/bin/ifortvars.sh ia32' and you must have installed the 32-bit compiler components, as well as the 32-bit g++.

You should note that all CPUs used in cluster computing which have been produced in the last 8 years are expected to perform better under x86_64 OS and several of the distros have cut back on support for 32-bit mode. Continuing to attempt support of 32-bit mode is conducive to the sort of confusion seen in this thread.