https://community.intel.com/t5/Intel-Moderncode-for-Parallel/Can-a-master-region-appear-in-OpenMP-work-sharing-region/m-p/992642#M6288 <DIV></DIV> <DIV>A question came up today about whether or not an OpenMP work-sharing region could contain a master region. There was no apparent language in the OpenMP Fortran spec that disallowed this (even in the latest 2.5 version). However, the compiler was not allowing this particular nesting of directives.</DIV> <DIV></DIV> <DIV>Anyone have any thoughts or advice on using a master region (code will only be executed by the master thread) inside a work-sharing construct (like parallel do)? Can you think of any algorithm that might need this functionality? Should this be allowed? If not, how would you try to "fix" the code that was written this way?</DIV> <DIV></DIV> <DIV>--clay</DIV> Fri, 08 Apr 2005 03:27:34 GMT ClayB 2005-04-08T03:27:34Z https://community.intel.com/t5/Intel-Moderncode-for-Parallel/Can-a-master-region-appear-in-OpenMP-work-sharing-region/m-p/992642#M6288 <DIV></DIV> <DIV>A question came up today about whether or not an OpenMP work-sharing region could contain a master region. There was no apparent language in the OpenMP Fortran spec that disallowed this (even in the latest 2.5 version). However, the compiler was not allowing this particular nesting of directives.</DIV> <DIV></DIV> <DIV>Anyone have any thoughts or advice on using a master region (code will only be executed by the master thread) inside a work-sharing construct (like parallel do)? Can you think of any algorithm that might need this functionality? Should this be allowed? If not, how would you try to "fix" the code that was written this way?</DIV> <DIV></DIV> <DIV>--clay</DIV> Fri, 08 Apr 2005 03:27:34 GMT https://community.intel.com/t5/Intel-Moderncode-for-Parallel/Can-a-master-region-appear-in-OpenMP-work-sharing-region/m-p/992642#M6288 ClayB 2005-04-08T03:27:34Z