The new home for Visual Studio documentation is Visual Studio 2017 Documentation on docs.microsoft.com.

Controls how loop code is to be considered by the auto-parallelizer, and/or excludes a loop from consideration by the auto-vectorizer.

#pragma loop( hint_parallel(n) )  

#pragma loop( no_vector )  

#pragma loop( ivdep )  


hint_parallel( n )
Hints to the compiler that this loop should be parallelized across n threads, where n is a positive integer literal or zero. If n is zero, the maximum number of threads is used at run time. This is a hint to the compiler, not a command, and there is no guarantee that the loop will be parallelized. If the loop has data dependencies, or structural issues—for example, the loop stores to a scalar that's used beyond the loop body—then the loop will not be parallelized.

The compiler ignores this option unless the /Qpar compiler switch is specified.

By default, the auto-vectorizer is on and will attempt to vectorize all loops that it evaluates as benefitting from it. Specify this pragma to disable the auto-vectorizer for the loop that follows it.

Hints to the compiler to ignore vector dependencies for this loop. Use this in conjunction with hint_parallel.

To use the loop pragma, place it immediately before—not in—a loop definition. The pragma takes effect for the scope of the loop that follows it. You can apply multiple pragmas to a loop, in any order, but you must state each one in a separate pragma statement.

Auto-Parallelization and Auto-Vectorization
Pragma Directives and the __Pragma Keyword