Text this: Are Large Language Models Smart Coders? Evaluating Correctness and Efficiency of LLM-generated Algorithms