GPU kernelsRapid Fire
GPU Kernel Fundamentals 1
Explain GPU kernel performance through seven short questions on memory access, thread execution and optimisation.
Unlock with Full AccessThe round
Work through CUDA concepts that affect how a kernel uses the GPU. The round covers performance limits, memory access and reuse, branching, synchronisation, occupancy and kernel fusion. You will explain mechanisms and make a short calculation using the stated setup.
- NVIDIA CUDA
- The questions use NVIDIA CUDA terminology. You can answer in prose without writing or running a kernel.
What you'll practise
Performance limits
Work out whether computation or memory bandwidth limits a kernel's throughput.
Memory access
Explain coalesced memory access and how shared memory supports data reuse.
Thread execution
Explain warp divergence, block synchronisation and how register use affects occupancy.
GPU kernel fundamentals
Answer seven questions
Get feedback on each answer when you finish the round.
All questions and feedback are included with Full Access.