 Effective classification algorithms are crucial for extracting valuable information from hyperspectral images, HSI, and ongoing research aims to improve upon existing algorithms. Due to the complexity of this problem and enormous computing time, parallel algorithms such as GPU accelerated computing have been developed to accelerate deep learning and other computationally intensive applications using general purpose graphics processing units, GPUs. This paper studies available GPU implementations for HSI classification, examining performance, major developments, and concerns in research work, and describing the challenges faced in GPU implementations for HSI. This article was authored by Aomig Yusuf and Shadia Lorne.