OpenX { X = MP/ACC } Directives

.
├── AccelerationToMulitcore
│   ├── 1.CPUOpenmp
│   ├── 2.SIMD
│   └── 3.AffinityOpenMP
├── AccelerationToGU
│   ├── GPUoffOpenACC
│   └── GPUoffOpenMP
├── Benchmarks
│   ├── OpenACC_ganga_reduction_collapse
│   ├── OpenACC_kernels_data_optimized
│   ├── OpenAcc_reduction_collapse
│   ├── OpenMP_reduction_collapse
│   ├── OpenMP_reduction_collapse_teams_gang
│   └── openmpPerformance
├── benchmarkfig1.html
└── benchmarkfig1.png

1. OpenMP Directives

2. Single instruction multiple data (SIMD)

3. OpenACC offloadings

   • Data movement
   • Loop parallelisms

4. OpenMP Offloadings

   • Target construct
   • The map clause
   • Target Teams construct
   • Nowait clauses
   • Distributed construct

5. Benchmark

Thanks to:

1. Tim Mattson (Intel)
2. Tom Deakin (University of Bristol)
3. Michael Klemm (AMD)
4. Jeff Larkin (Nvidia)
5. Manuel Arenaz (NERSC)
6. Luca and Sergio (CINECA) and many others OpenMP 5.0.1 specification and examples https://www.openmp.org/resources/