Parallelization option - static vs. dynamic arrays

[wasif1]

We have a program that we compiled in a Windows XP (x64, 4 processors) workstation using Fortran 9.0 with Visual Studio 2003. This program used some relatively large static arrays, but not to exceed the 2GB limit. Let's call it PROGRAM 1.

Then, we converted the program to use larger dynamic (allocatable) arrays. We compiled the new version of the program with the trial version of Fortran 9.1 and Visual Studio 2005. Let's call it PROGRAM 2.

Both programs were compiled using the "parallelization" option in the compiler.

According to the person who is testing the program, PROGRAM 1 shows good parallelization (you can see the four processors active), while PROGRAM 2 only shows one processor active.

We're getting the full professional version of Fortran 9.1 today and will be testing PROGRAM 2 again with this version, however, in the mean time, we would like to check about the following issues:

1) Does parallelization works differently if you have static vs. dynamic arrays?

2) Are there some settings in Visual Studio 2005 that need to be tweaked to allow for full parallelization?

3) Where can we find a good description of the parallelization option in the Fortran manual.

Thanks

[Steven_L_Intel1]

We supply an "Optimizing Applications" manual along with the other on-disk documentation which discusses parallel programming in detail. There is no difference between the Standard and Professional Editions in this regard.

I urge you to read the relevant sections of the Optimizing Applications manual. Getting good parallel speedup is not just a matter of throwing a switch or tweaking settings.