基于模糊模型的GPU性能评估研究

传统的GPU性能评估一般都采用定性评估方法，难以在保证效率的情况下给出一个准确的评估结果。介绍了确定影响GPU性能的关键指标，并给出了对GPU性能评估的理论基础，应用模糊数学中的模糊综合评判方法建立数学模型，最终运用该模型得出了GPU性能的定量评估值。实验表明，方法为准确判断GPU性能提供了科学的依据，且具有较好的实用性和可靠性。

Research on GPU Performance Evaluation Based on Fuzzy Model

Generally, the traditional performance evaluation of GPU uses the qualitative approach, it is difficult to give a precise evaluation result under guaranteeing efficiency. This paper introduces an accurate quantitative evaluation method aiming to provide a scientific basis in the process of evaluating the GPU performance by applying the following methods. First, it is crucial to identify the key factors putting impact on the GPU performance. At the same time, a theoretic basis is provided for the GPU evaluation. Then, a mathematical model is generated based on fuzzy comprehensive evaluation method in the fuzzy math. Finally, a quantitative evaluation result of GPU performance can be acquired by applying the mathematical model. Meanwhile, it has excellent practicability and reliability.