each_WinNNpar.EXE_requires_one_DLL.txt:

The files in this folder may include up to 4 executable (EXE) files for Windows:
   [name]-Win32seq.exe, sequential version for 32-bit Windows (also known as the "IA-32" platform);
   [name]-Win32par.exe, parallel   version for 32-bit Windows (also known as the "IA-32" platform);
   [name]-Win64seq.exe, sequential version for 64-bit Windows (also known as the "x64"   platform).
   [name]-Win64par.exe, parallel   version for 64-bit Windows (also known as the "x64"   platform).
Each of this executables was compiled using Intel's "Parallel Studio XE 2013"
   Fortran compiler, linking-in their Intel Math Kernel Library (MKL) software
   for numerical tasks.  For "par" executables, compiler & linker switches were set to
   provide PARALLEL computing (i.e., multi-threading, even on a single processor chip;
   or up to 16-fold parallel processing if you have two high-end CPU chips)
   to speed up certain time-consuming task(s), especially when calling MKL.
However, each of the -WinNNpar.exe files requires access to the appropriate version
   (32-bit or 64-bit) of Intel's redistributable DLL (dynamic link library)
   file "libiomp5md.dll".  If you try to run a -WinNNpar.EXE on Windows without
   having the necessary DLL file available, you will get an error message and the
   program will stop, immediately.  (This does not cause any harm, however.)
Please see my top-level folder, "REQUIRED_libiomp5md.dll_file" to get the
   appropriate DLL file for your system.  There are also more text files
   in that folder which will advise you on where to store the DLL file.
   The short answer is that this DLL file should usually be placed in the same
   folder where you put input data files required by the program; this may also
   be the same folder in which you store the -WinNNpar.exe file(?).
If you are wondering whether any parallel processing is actually occurring when
   you run these codes, the Windows utility "Task Manager" has a "Performance"
   tab that can display the activity of each of the processors on your CPU chip(s).
   Please understand that only certain portions of the code will/may run in
   parallel mode.  These are the portions that are executed in MKL, and perhaps
   a few more loops that I have had the time/energy/skill to coax into parallel.

Peter Bird
UCLA
2015.08.31