GPU Memory Management and Fragmentation Quiz

Answers and explanations appear as you take the quiz.

1. What is GPU memory fragmentation?
   • When the GPU physically breaks apart
   • When the model has too many layers
   • When two GPUs disagree on a result
   • When free memory is split into scattered chunks too small to satisfy a large contiguous request
2. Why do frameworks like PyTorch use a caching memory allocator?
   • To make the model more accurate
   • To avoid slow direct driver allocation/free calls by reusing pooled memory blocks
   • To encrypt the weights
   • To reduce the number of GPUs needed to one
3. A CUDA allocation fails despite apparently unused reserved memory. What should a developer investigate?
   • Whether fragmentation is proven by the error alone
   • Whether all live tensors have already been freed
   • Whether inactive split blocks or other allocations explain the failure
   • Whether changing the learning rate releases GPU memory
4. Which PyTorch allocator option can be tested for changing-size workloads that leave unused memory slivers?
   • Expandable segments in the allocator configuration
   • Increasing every tensor dimension
   • Disabling the GPU cooling system
   • Increasing the optimizer learning rate
5. What does vLLM's PagedAttention manage to reduce memory waste during inference?
   • The model's training data
   • The attention key-value (KV) cache, using fixed-size pages
   • The GPU's clock speed
   • The font size of logs
6. Which technique reduces activation memory to help avoid out-of-memory errors?
   • Adding more layers
   • Using a longer learning-rate warmup
   • Gradient checkpointing (recomputation)
   • Saving more checkpoints to disk
7. A live CUDA tensor still has Python references. Will empty_cache release that tensor’s allocation?
   • No; it releases unused cached memory rather than live tensor storage
   • Yes; it deletes every referenced tensor
   • Yes; it moves the tensor automatically to NVMe
   • Only if the tensor contains zero values
8. PyTorch reports reserved memory greater than allocated memory. What distinction explains the gap?
   • Reserved memory measures only CPU RAM
   • Allocated memory includes every process on the GPU
   • The difference directly proves a model accuracy error
   • The allocator can retain cached blocks beyond memory currently used by tensors