OpenMP failure

[Andrew_Smith]

I have a large section of code surrounded by an !$OMP PARALLEL block. When the execution hits the top of the block it jumps right to the end! No worker threads are created and all the code execution is skipped. Another similarly constructed block in my same application works just fine.

Is this a compiler bug?

[Steven_L_Intel1]

It is impossible to know without a test case. Please provide one and we'll be glad to investigate.

[jimdempseyatthecove]

What happens when you compile that file with OpenMP disabled?

Have you verified that you do not exit the parallel region by any way other than (all threads) falling through the !$OMP END PARALLEL

i.e. you cannot escape the region with a GOTO to a label outside of the region or RETURN or error transfer label on READ, etc...

(error message should be displayed, maybe it did not)

What version is your compiler. Some of the older ones (e.g. V8.n.n and V9.n.n) had problems with very large !$OMP blocks.

Also, is this compiled in Release build (optimizations cause the single step to walk your code in wierd order).

Jim Dempsey