OpenMP user-defined reductions


I can tell that nvc++ doesn’t support #pragma omp declare reduction(...) by the error I get:

error: this OpenMP construct is not supported in NVIDIA subset: declare reduction for user defined reductions

Are you planning to support that?

I’m using mp=multicore with the following version:

nvc++ 23.5-0 64-bit target on x86-64 Linux -tp haswell
NVIDIA Compilers and Tools
Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES.  All rights reserved.


Hi Christos,

Currently our OpenMP team is working on hardening our offload implementation. Hopefully they’ll be able to start adding additional features such as user defined reductions in the relatively near future, but we don’t have a specific timeline on when this would occur.


Hi Mat,

That’s good to know, thanks.