Introduction to OpenMP Offloading¶

1. In OpenMP Offloading, which directive is primarily used to specify code regions that should be executed on a device, such as a GPU?

• A. #pragma omp parallel
• B. #pragma omp for
• C. #pragma omp target
• D. #pragma omp simd

Device Memory Management and Data Movement¶

2. Which of the following map clauses is used to transfer data from the host (CPU) to the device (GPU) before computation?

• A. map(from:)
• B. map(to:)
• C. map(tofrom:)
• D. map(alloc:)

3. True or False: In OpenMP Offloading, the map(tofrom:) clause ensures that data is updated both on the device and on the host, allowing bidirectional data movement.

Offloading and Parallelization¶

4. Which directive divides loop iterations across multiple teams on a device, making it particularly useful for GPU offloading in OpenMP?

• A. teams
• B. declare target
• C. distribute
• D. target

Compilation and Tool Support¶

5. To enable OpenMP Offloading targeting an NVIDIA GPU using Clang, which compilation flag would you use?

• A. -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa
• B. -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda
• C. -mp -gpu=cc70
• D. -fopenmp-targets=spir64

Practical Understanding¶

6. True or False: OpenMP Offloading's teams and parallel directives can be combined to organize work distribution across teams of threads on a GPU.