Lets you select between old or new implementations of parallel loop support.
Linux and macOS:
- Specifies which implementation to use. Possible values are:
- Enables the new implementation of parallel-loop support. As a result, parallel C++ range-based loops and collapsing complex loop stacks will not result in compilation errors. This is the default.
- Enables the old implementation of parallel-loop support. This is the same implementation that was supported in 18.0 and earlier releases.
- This is the same as specifyingnew.
- The compiler uses the new implementation of parallel-loop support. Note that this setting may not yet be as stable as setting "old" since the implementation is new.
This option lets you select between old or new implementations of parallel loop support.
The new implementation handles parallel C++ range-based loops
, and also collapsing of OpenMP* parallel loops with complicated bounds expressions,for which the previous implementation reported errors.
If your code has a parallel loop that is not handled by the previous implementation, we recommend that you enable use of the new implementation.