| title | ms.date | f1_keywords | ms.assetid | ||
|---|---|---|---|---|---|
loop pragma |
08/29/2019 |
|
6d5bb428-cead-47e7-941d-7513bbb162c7 |
Controls how loop code is to be considered by the auto-parallelizer, 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 )
A hint 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. It's a hint to the compiler, not a command. There's no guarantee that the loop will be parallelized. If the loop has data dependencies, or structural issues, then it won't be parallelized. For example, it isn't parallelized if it stores to a scalar that's used beyond the loop body.
The compiler ignores this option unless the /Qpar compiler switch is specified.
no_vector
By default, the auto-vectorizer attempts to vectorize all loops that it evaluates may benefit from it. Specify this pragma to disable the auto-vectorizer for the loop that follows.
ivdep
A hint to the compiler to ignore vector dependencies for this loop.
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