You can specify the compute capability you are targeting with the option -gpu=ccXX, (V100 will be cc70, A100 will be cc80, H100 will be cc90). So, in your case it will be -gpu=cc90.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Behaviour of OpenMP target maps with Fortran arrays | 12 | 147 | February 11, 2025 | |
| Fortran OpenMP offloading painfully slow on NVIDIA architectures | 1 | 33 | July 28, 2025 | |
| Out of range error with openmp gpu offload | 10 | 1026 | February 1, 2023 | |
| OpenMP Multi-GPU, not getting speedup expected | 5 | 5866 | July 15, 2011 | |
| CUDA & openMP Problem with the SDK sample code | 11 | 14041 | September 12, 2015 | |
| accelerate a single loop with mpi and gpu | 21 | 15952 | July 19, 2013 | |
| Different GPU memory usage between OpenACC and OpenMP Offload | 10 | 909 | April 28, 2023 | |
| Questions about omp offload and memory transfer | 13 | 1489 | October 15, 2021 | |
| Running nvidia Fortran on multiple GPUs with MPI | 36 | 454 | December 14, 2024 | |
| NVFORTRAN-F-0000-Internal compiler error. gen_llvm_expr(): no incoming ili 0 | 7 | 41 | December 25, 2024 |