void foo(a,b,c)
   int a;
   float b;
   float c[10];
{
   int i = 0;
   for(i=0;i<10;i++)
      c[i]=b*a;
}

• a module-definition (.def) file
• __declspec(dllexport)

The Intel® C++ Compiler for Windows supports both methods.

But please note on how the .def is supported.
The module-definition (.def) file is ONLY used at link time for both Intel C++ compiler and Visual C++. When /Qipo (/GL) is used, the Intel C++ compiler will do more aggressive inlining and optimization. Some functions may be completely inlined so the function body will be removed at compile time if there is no "__declspec(dllexport)" in front of the function.
In such cases if the function name is in the .def file, you will get a link error "xxx.def : error LNK2001: Unresolved external symbol "YYY"".

This is not a bug in the Intel C++ compiler for Windows.

The resolution for this:

• add the "__declspec(dllexport)" in front of the function declaration so the Intel C++ compiler will know at compile time that this function will be used for exporting.

Problem :  gcc and the Intel® C++ Compiler* provide macros that are defined when the -msse (gcc) or -x (Intel) options are used.  The Intel® C++ Compilers 11.0.069 for Linux*,  11.0.054 for Mac OS* and 11.0.061 for Windows* support -xSSE4.1 and -xSSE4.2 options, but do not provide the corresponding __SSE4_1__ and __SSE4_2__ macro definitions.


Environment : Windows*, Linux* or Mac OS* X; C++; IA-32 or Intel® 64 platforms


Root Cause : Already stated above


Resolution : 
These macros are now defined in

• the Intel® C++ Compilers for Linux* 11.0.074 and later
• the Intel® C++ Compilers for Mac OS* 11.0.056 and later
• the Intel® C++ Compilers for Windows* 11.0.066 and later
• the Intel Parallel Composer

IA-32 Architecture refers to systems based on 32-bit processors generally compatible with the Intel Pentium® II processor, (for example, Intel® Pentium® 4 processor or Intel® Xeon® processor), or processors from other manufacturers supporting the same instruction set, running a 32-bit operating system.

Intel® 64 Architecture (formerly Intel® EM64T)refers to systems based on IA-32 architecture processors which have 64-bit architectural extensions, (for example, Intel® Core™2 processor family), running a 64-bit operating system such as Microsoft Windows Vista* x64 or a Linux* "x86_64" variant. If the system is running a 32-bit  operating system, then IA-32 architecture applies instead. Systems based on AMD* processors running a 64-bit operating system are also supported by Intel compilers for Intel® 64 architecture applications.

64-bit computing on Intel architecture requires a computer system with a processor, chipset, BIOS, operating system, device drivers and applications enabled for Intel® 64 architecture. Performance will vary depending on your hardware and software configurations. Consult with your system vendor for more information.

IA-64 Architecture refers to systems based on the Intel® Itanium® processor running a 64-bit operating system.

libiomp5md.dll, libiomp5md.lib: dynamic performance library; options "/MD /Qopenmp-link:dynamic"

libiomp5mt.lib: static performance library; options "/MT /Qopenmp-link:static"

libiompprof5md.dll, libiompprof5md.lib: dynamic profile library; options "/MD /Qopenmp-profile"

libiompprof5mt.lib: static profile library; options "/MT /Qopenmp-profile"

libiompstubs5md.dll, libiompstubs5md.lib: dynamic stub library; options "/MD /Qopenmp-stubs"

libiompstubs5mt.lib: static stub library; options "/MT /Qopenmp-stubs"

Visual C++ 2005* provides only dynamic OpenMP libraries: vcomp.lib, vcomp.dll
Visual C++ 2008* provides only dynamic OpenMP libraries: vcomp90.lib, vcomp90.dll

In one application there should be only ONE OpenMP run time library.

The Intel OpenMP compatibility libraries can be used with Microsoft* Visual C++ 2005 and 2008*. Followings are some examples on how to use:

• Using the Intel Compiler OpenMP compatibility libraries with Visual C++ 2008:

>> cl /c /MD /D_OPENMP_NOFORCE_MANIFEST /openmp ompSrc1.cpp
>> link /nodefaultlib:vcomp90 libiomp5md.lib ompSrc1.obj

• Using the Intel Compiler OpenMP compatibility libraries with both Visual C++ 2008 and Intel C++ Compiler:

>> cl /c /MD /D_OPENMP_NOFORCE_MANIFEST /openmp ompSrc1.cpp
>> icl /c /MD /Qopenmp ompSrc2.cpp
>> xilink /nodefaultlib:vcomp90 libiomp5md.lib ompSrc1.obj ompSrc2.obj

The legacy OpenMP runtime library
The legacy OpenMP runtime library is deprecated in 11.x release. In case you're using a 3rd party library that uses the legacy OpenMP runtime, you can use the following option to link with the legacy libraries:

/Qopenmp-lib:legacy (note: deprecated)

The primary files are below: for full list of files, please see the Intel C++ or Fortran Compiler Documentation.

libguide40.dll, libguide40.lib: dynamic library

libguide.lib: static library

But the Intel(R) C++ Compiler in the Intel(R) Parallel Composer beta has significantly reduced the number of diagnostic messages emitted by default; and it tries to match the diagnostic number emitted by Visual C++ as much as possible. So when you disable a warning in the code or by option, the warning should be disable when you use Intel C++ compiler.

We have an on-going project to document all the diagnostic messages emitted by the Intel C++ Compiler. There will be a Knowledge Base article for each diagnostic that will eventually contain detailed information about the diagnostic as well as examples. Right now all of the diagnostic articles are already published except only a few. If you'd like to see detailed information for a particular diagnostic, please send us a request on the diagnostic article you're interested.

To find all the diagnostic articles, visit http://software.intel.com/en-us/articles/intel-parallel-composer/all/1 and click "Error Message".