Intel® Fortran Compiler 16.0 User and Reference Guide
To enable the auto-parallelizer, use the [Q]parallel option. This option detects parallel loops capable of being executed safely in parallel, and automatically generates multi-threaded code for these loops.
You may need to set the KMP_STACKSIZE environment variable to an appropriately large size to enable parallelization with this option.
Using this option enables parallelization for both Intel® microprocessors and non-Intel microprocessors. The resulting executable may get additional performance gain on Intel® microprocessors than on non-Intel microprocessors. The parallelization can also be affected by certain options, such as /arch (Windows*), -m (Linux* and OS X*), or [Q]x.
An example of the command using auto-parallelization is as follows:
|
Commanding auto-parallelization in Linux* |
|---|
ifort -c -parallel myprog.f |
|
Commanding auto-parallelization in Windows* |
|---|
ifort -c /Qparallel myprog.f |
|
Commanding auto-parallelization in OS X* |
|---|
ifort -c -parallel myprog.f |
Auto-parallelization uses two specific directives: !DIR$ PARALLEL and !DIR$ NOPARALLEL.
The format of an auto-parallelization compiler directive is below:
|
Syntax |
|---|
!DIR$ <directive> |
Because auto-parallelization directives begin with an exclamation point, the directives take the form of comments if you omit the [Q]parallel option.
The !DIR$ PARALLEL directive instructs the compiler to ignore dependencies that it assumes may exist and that would prevent correct parallelization in the immediately following loop. However, if dependencies are proven, they are not ignored. In addition, PARALLEL [ALWAYS] overrides the compiler heuristics that estimate the likelihood that parallelization of a loop increases performance. It allows a loop to be parallelized even if the compiler thinks parallelization may not improve performance. If the ASSERT keyword is added, as in !DIR$ PARALLEL [ALWAYS [ASSERT]], the compiler generates an error-level assertion message saying that the compiler analysis and cost model indicate that the loop cannot be parallelized.
The !DIR$ NOPARALLEL directive disables auto-parallelization.
For example, in the following code, the NOPARALLEL directive disables auto-parallelization.
|
Example |
|---|
program main parameter (n=100 integer x(n),a(n) !DIR$ NOPARALLEL do i=1,n x(i) = i enddo !DIR$ PARALLEL do i=1,n a( x(i) ) = i enddo end |