Matrix Multiplication¶

1. In the parallelized version of matrix multiplication using OpenACC, which clause is used to combine nested loops for simultaneous parallelization?

• A. tile
• B. collapse
• C. copyin
• D. async

Data Clauses and Memory Allocation¶

2. True or False: The copyin clause in OpenACC transfers input data from the host (CPU) to the device (GPU) before computation begins.

Reduction and Optimization Clauses¶

3. Which OpenACC clause is used to ensure safe accumulation of the sum variable across parallel threads in matrix multiplication?

• A. tile
• B. loop
• C. reduction
• D. present

4. True or False: The tile clause in OpenACC breaks down the iteration space into smaller blocks, which can help reduce memory latency on the GPU.

Fortran and C/C++ Differences¶

5. When using OpenACC in Fortran, why is array indexing adjusted by +1 in single-dimensional storage?

• A. To align with C/C++ indexing
• B. For compatibility with OpenACC
• C. Because Fortran arrays start indexing from 1 by default
• D. To avoid memory conflicts