Intel® Fortran Compiler for Linux* and Mac OS X*

Intel Ifort 15 + OpenMPI 1.8.4 + OpenMP= instantaneous segfault

Hi,

using Intel ifort 15, I compiled OpenMPI 1.8.4 using the following configure line:

../configure --prefix=<path to installdir>  --with-openib --with-sge CC=icc FC=ifort CXX=icpc

Unfortunately, compiling our hybrid MPI + OpenMP code with the resulting MPI compiler wrapper results in a binary which segfaults instantaneously after startup.

Please consider the following example which shows the same behavior as our large code:

compilervars.sh destabilizes my linux environment

 

I have tried to install Intel Composer XE 2015.0.090 in both OpenSuse 13.1 and Ubuntu 14.04 LTS environments. In both cases I have followed the instructions here (https://software.intel.com/en-us/articles/using-intel-compilers-for-linu...)  and installed the compilers for 64bit system only.

When I add the

source /opt/intel/bin/compilervars.sh intel64 

Incorrect result with COUNT in initialization expression

I found that Intel Fortran 15 gives incorrect results with the COUNT intrinsic in initialization expressions. Consider the following program:

program count_bug
  implicit none

  integer, parameter :: npts(2) = (/ 42, 1 /)
  integer, parameter :: ndims = count(npts .gt. 1)

  write (*,'(2(I0, 1X))') count(npts .gt. 1), ndims
end program

When compiled with ifort, the output is "1 -1".  The correct output is "1 1".

This result was obtained with Intel Fortran 15.0.1 20141022 on Mac OS X 10.10.1.

Convenience libraries, ipo, and unresolved references

I am attempting to recompile code that was written using GNU tools, including autotools. In the makefiles, I make a lot of use of so-called convenience libraries, or local static libraries that are only used as compilation stages, and which are not installed. This is a fairly common practice with autotools, but doing so leads to unresolved references at link time when -ipo is specified on the command line. If I don't use -ipo, everything is fine. The problem only occurs with -ipo. 

Is there an easy workaround that does not involve removing the convenience libraries? 

vtune_amplifier_xe_2013 + how to compile

dear all,

I have vtune_amplifier_xe_2013, I used it one year ago to analyze the CPU time in my program. 

I remember that it produce the files: .dump and .xml

I do not remember anymore how to compile the program to get the previous files.

I do not remember the flags that I have to use in ifort.

Someone could help me, please?

I am not able to find the guide anymore. Now I am trying to look inside the  vtune_amplifier_xe_2013 folder.

Thanks

 

Question: Reading a large column (1D-array) and writing slices of this array as rows

I am trying to read a large single column (which I am trying to read in as a 1-D array), and then to write slices of this array as rows until the end of the data. Please suggest me the best way to accomplish this task. Thanks in advance.

Have input variables from header file be recognized as initialized variables

Hey,

I am currently debugging my code for a runtime error (invalid pointer) but the compiler does not give me valuable information for locating this error even with -traceback. I am therefore make use of all the debug flags I found online with the hope that it may solve the problem:

FLAGS = -O2 -stand f03 -std -check all -traceback -warn all -fstack-protector -assume protect_parens -implicitnone

 

Using these flags, I actually get a lot more error than the one I had to begin with. One of them is about a variable name which does not have a type:

Customizing Makefile for the programm using Fortran Compiler.

Hi!

I'm trying to define the makefile for compiling the programm in Linux that uses Fortran Compiler. In the programm manual it is written that it requires FORTRAN90 compiler and MPI. I've installed Intel Parallel Sdudio 2015 and openMPI. The target template in the makefile looks like this:

订阅 Intel® Fortran Compiler for Linux* and Mac OS X*