1. What does CUDA stand for?
A. Central Unit for Data Access
B. Compute Unified Device Architecture
C. Core Unit Data Application
D. Computer Use Direct Access
Answer: B
2. CUDA is mainly used for:
A. Web development
B. Graphics designing
C. Parallel computing on GPU
D. Database management
Answer: C
3. CUDA C++ is based on which language?
A. Java
B. Python
C. C++
D. Ruby
Answer: C
4. The __global__ keyword is used to define:
A. CPU function
B. File structure
C. Device kernel
D. Variable scope
Answer: C
5. CUDA kernels are launched using:
A. Curly braces {}
B. Square brackets []
C. Triple angle brackets <<< >>>
D. Round brackets ()
Answer: C
6. In CUDA, the host is:
A. GPU
B. RAM
C. CPU
D. SSD
Answer: C
7. In CUDA, what is a 'block'?
A. A storage unit
B. A group of threads
C. A GPU memory part
D. A GPU chip
Answer: B
8. A 'grid' in CUDA is made up of:
A. Registers
B. Threads
C. Blocks
D. Kernels
Answer: C
9. What is the index of the first element in CUDA arrays?
A. 1
B. 0
C. -1
D. Depends on size
Answer: B
10. Which variable is used to access the current block index?
A. block.x
B. blockDim.x
C. blockIdx.x
D. blockIndex
Answer: C
11. Which variable is used to access the current thread index?
A. thread.x
B. threadIndex
C. blockDim
D. threadIdx.x
Answer: D
12. Which CUDA function allocates managed memory?
A. cudaAlloc()
B. malloc()
C. cudaMallocManaged()
D. new()
Answer: C
13. Which CUDA function frees allocated memory?
A. delete()
B. cudaFree()
C. free()
D. cudaRemove()
Answer: B
14. What does the following line do? add<<<1, N>>>();
A. Launches 1 thread
B. Launches N blocks
C. Launches 1 block with N threads
D. Compiles the code
Answer: C
15. What does blockDim.x represent?
A. Block index
B. Thread count per block
C. Grid size
D. Kernel size
Answer: B
16. What is the formula to calculate thread index across blocks?
A. blockIdx.x * threadIdx.x
B. blockIdx.x + threadIdx.x
C. threadIdx.x * blockDim.x
D. threadIdx.x + blockIdx.x * blockDim.x
Answer: D
17. What is used to avoid accessing beyond the end of an array?
A. A break statement
B. An if condition
C. A continue statement
D. A while loop
Answer: B
18. In CUDA, threads can:
A. Only run independently
B. Not share data
C. Communicate and synchronize
D. Only read memory
Answer: C
19. What is the advantage of combining blocks and threads?
A. Better image quality
B. Simpler code
C. More parallelism
D. Uses less memory
Answer: C
20. CUDA threads are grouped under:
A. Loops
B. Blocks
C. Registers
D. Functions
Answer: B
21. Which of the following is a built-in CUDA variable?
A. mainIndex
B. threadCount
C. threadIdx.x
D. blockCount
Answer: C
22. What is the purpose of random_ints() in CUDA code?
A. Allocate memory
B. Free memory
C. Fill arrays with random values
D. Compile the kernel
Answer: C
23. Why do we use (N + M - 1) / M in kernel launch?
A. To waste GPU time
B. To make kernel launch simpler
C. To handle any array size
D. To reduce memory usage
Answer: C
