cuda_unified_memory_mpi_bcast

Run

$ mpirun -n 2 --allow-run-as-root ./cuda_unified_memory_mpi_bcast

Reference

• Unified Memory for CUDA Beginners
• Unified Memory: Now for CUDA Fortran Programmers
• Combine OpenACC and Unified Memory for Productivity and Performance
• Maximizing Unified Memory Performance in CUDA

Legacy:

• Unified Memory in CUDA 6
• Beyond GPU Memory Limits with Unified Memory on Pascal