Abstract It is a phenomenon well-known to professional and private users of complex machines: in most situations these machines do not deliver their full performance. This difference between the actual achieved performance and some maximal performance obtained in benchmarks varies between different use cases but poses crucial questions concerning the resulting gap in any case. Is it possible to reduce this manifest inefficiency by changing the machine design or the process characteristics in order to increase end user satisfaction and confirm the rationale of the machine investment? In this article we present a novel systematic derivation and compilation of the causes of this performance loss based on geometric and stochastic models that may help in analysing and optimising the average performance of complex machines in real-world scenarios.

中文翻译：

---

确定机器性能损失的数学模型

摘要 这是复杂机器的专业和私人用户所熟知的一种现象：在大多数情况下，这些机器无法发挥其全部性能。实际实现的性能与基准测试中获得的某些最大性能之间的差异在不同的用例之间有所不同，但在任何情况下都提出了有关由此产生的差距的关键问题。是否可以通过改变机器设计或过程特性来减少这种明显的低效率，以提高最终用户的满意度并确认机器投资的基本原理？在本文中，我们基于几何和随机模型提出了一种新颖的系统推导和对这种性能损失原因的汇编，这可能有助于分析和优化现实世界场景中复杂机器的平均性能。