NCA CUDA Programming & Parallel Computing

Question 1

What programming language is commonly used for CUDA development?

Accepted Answer

CUDA C/C++

Answer

Python only.

Answer

Java.

Answer

Fortran.

Question 2

Which CUDA memory type is shared between threads in the same block?

Accepted Answer

Shared memory

Answer

Global memory.

Answer

Constant memory.

Answer

Register memory.

Question 3

What is a CUDA kernel?

Accepted Answer

A GPU function executed in parallel

Answer

A CPU function.

Answer

A memory allocation routine.

Answer

A graphics shader.

Question 4

What does occupancy mean in CUDA programming?

Accepted Answer

Ratio of active warps to max warps

Answer

The percentage of GPU clock speed.

Answer

The amount of memory used.

Answer

Number of threads per block.

Question 5

Which CUDA API function launches kernels?

Accepted Answer

kernel<<<grid, block>>>

Answer

cudaLaunch().

Answer

cudaStart().

Answer

kernelLaunch() function.

Question 6

What is warp in CUDA terminology?

Accepted Answer

Group of 32 threads

Answer

A single thread.

Answer

A GPU core.

Answer

A block of memory.

Question 7

Which type of memory has the highest latency in CUDA?

Accepted Answer

Global memory

Answer

Register memory.

Answer

Shared memory.

Answer

Texture memory.

Question 8

What does CUDA streams enable?

Accepted Answer

Concurrent execution

Answer

Sequential execution only.

Answer

Memory allocation.

Answer

Kernel compilation.

Question 9

How does CUDA handle synchronization between threads?

Accepted Answer

Using __syncthreads()

Answer

Using mutexes only.

Answer

Threads run independently.

Answer

No synchronization is possible.