While SHA-2 is a ubiquitous cryptographic hashing primitive, its role in emerging application domains, such as blockchains or trusted IoT components, has made the acceleration of SHA-2 very challenging due to new stringent classes of requirements imposed by such domains, especially implementation cost and energy efficiency. This Ph.D. thesis explores the SHA-2 design space from different viewpoints. Its first contribution is a reasoned classification of the many SHA-2 designs proposed in the literature according to their architectural choices, each of them having different implications on the application requirement. Based on this analysis, this thesis introduces a framework and a methodology for evaluating and comparing different implementation options, which is used to assess the impact of each architectural technique on the application requirements, as well as the effect of variations in the underlying target technology. The last contribution of this thesis explores a different approach, namely utilising a specific target technology with maximum efficiency, and the resulting SHA-2 accelerator shows the best area efficiency reported so far in the literature.