HPC-Quiz-FT IV

What is the correct way to free allocated GPU memory?

• free(ptr);

• cudaFree(ptr);

• gpuFree(ptr);

• cudaDealloc(ptr);

Which function is used to copy data from the CPU to the GPU?

• cudaMemcpy(dest, src, size, cudaHostToDevice);

• cudaCopy(dest, src, size, cudaMemcpyHostToDevice);

• cudaMemcpy(dest, src, size, cudaMemcpyHostToDevice);

• memcpy(dest, src, size);

How do you allocate memory on the GPU?

• malloc(size);

• cudaMalloc(&ptr, size);

• gpuAlloc(&ptr, size);

• cudaAlloc(&ptr, size);

What does the threadIdx.x variable represent?

• The index of the current thread within a block

• The index of the current block within the grid

• The total number of threads in a block

• The total number of blocks in the grid

Which keyword is used to define a function that runs on the GPU but is called from the CPU?

• __device__

• __global__

• __host__

• __gpu__

How do you declare a CUDA kernel function?

• void kernel<<<...>>>();

• __global__ void kernel() {}

• __kernel void kernel() {}

• gpu_function void kernel() {}

How do you launch a CUDA kernel with 256 threads per block?

• kernel();

• kernel<<<1, 256>>>();

• kernel<<<256, 1>>>();

• kernel<<<256>>>();

Which company developed CUDA?

• Intel

• AMD

• NVIDIA

• Qualcomm

What is the purpose of CUDA streams?

• To manage parallel kernel execution

• To synchronize CPU and GPU execution

• To store global memory

• To create new CUDA threads

What is the maximum number of threads in a CUDA block (typically)?

• 128

• 512

• 1024

• 2048