INNER CODE UNIT · Python
get_gpu_stats
ai-infra-curriculum/ai-infra-engineer-learning · projects/project-103-llm-deployment/src/llm/server.py:297
def get_gpu_stats(self) -> Dict[str, float]:
"""
Get current GPU utilization statistics.
Returns:
Dictionary with GPU metrics
TODO: Implement GPU monitoring:
1. Query CUDA memory usage (allocated, cached, free)
2. Get GPU utilization percentage
3. Get GPU temperature
4. Calculate KV cache usage
Hints:
- Use torch.cuda.memory_allocated()
- Use torch.cuda.memory_reserved()
- Use pynvml for detailed GPU stats
"""