Heterogeneous Computing: Integrating CPUs, GPUs, TPUs, and FPGAs for High-Performance AI Processing
Year: 2019
Type: article
Abstract: In parallel computing, we have a way of experiencing the growth of computational requirements of Artificial Intelligence (AI) and Machine Learning (ML) procedures known as heterogeneous computing. Standard computation structures, like CPUs, have failed to support large-scale computations for AI, thus the integration of other computation processors, including GPUs, TPUs, and FPGAs. This paper discusses the advantages and disadvantages of having heterogeneous computing architecture, especially in AI and deep learning systems. Combining SoC processor types can enhance High-performance AI surgery's effectiveness, adaptability, and energy use. We compare different architectures, describe practical solutions, and focus on various approaches to resource management. Additionally, we discuss the examples of AI in specific areas like picture and pattern recognition, NLP, and self-driving systems. By using the information obtained in this paper, it is possible to conclude that heterogeneous computing is a promising direction in the evolution of AI and to draw some expectations regarding further advancements in this field.
Author: Karthik Wali
Language: English
Subfield: Hardware and Architecture
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4411570728