Architectural considerations for FPGA acceleration of machine learning applications in MapReduce
Katayoun Neshatpour, Hosein Mohammadi Mokrani, Avesta Sasan, Hassan G Ghasemzadeh, Setareh Rafatirad, Houman Homayoun · 2018
While demand for data center computational resources continues to grow as the size of data grows, the semiconductor industry has reached scaling limits and is no longer able to reduce power consumption in new chips. Unfortunately, the promise of analytics running on large systems (e.g., data-centers) over huge data, and scaling those systems into the future, coincides with an era when the end of Dennard scaling brings into serious question our ability to provide scalable computational power without prohibitive power and energy costs.