Key CUDA and HIP APIs and Qualifiers¶

1. Which function qualifier in CUDA and HIP is used to define a function that executes on the GPU and is callable from the CPU?

• A. __host__
• B. __device__
• C. __global__
• D. __constant__

Variable Qualifiers in CUDA and HIP¶

2. True or False: The __shared__ qualifier in CUDA and HIP specifies that the variable is accessible only by threads within the same block.

3. Which qualifier is used to define a variable that resides in read-only memory and is accessible by all threads on the GPU?

• A. __device__
• B. __shared__
• C. __global__
• D. __constant__

Thread Qualifiers in CUDA and HIP¶

4. What does threadIdx represent in CUDA and HIP?

• A. The number of threads in each block.
• B. The index of the current thread within its block.
• C. The dimensions of the grid.
• D. The number of blocks in the grid.

5. In CUDA, a warp is composed of how many threads?

• A. 16
• B. 32
• C. 64
• D. 128

6. True or False: In HIP (for AMD GPUs), a wavefront typically contains 64 threads.

Memory Access in CUDA and HIP¶

7. Which memory qualifier would you use to store a variable in global memory, making it accessible from both the host and the GPU?

• A. __device__
• B. __shared__
• C. __constant__
• D. __local__