GANPU: A 135TFLOPS/W Multi-DNN Training Processor for GANs with Speculative Dual-Sparsity Exploitation

Abstract

Generative adversarial networks (GAN) have a wide range of applications, from image style transfer to synthetic voice generation [1]. GAN applications on mobile devices, such as face-to-Emoji conversion and super-resolution imaging, enable more engaging user interaction. As shown in Fig. 7.4.1, a GAN consists of 2 competing deep neural networks (DNN): a generator and a discriminator. The discriminator is trained, while the generator is fixed, to distinguish whether the generated image is real or fake. On the other hand, the generator is trained to generate fake images to fool the discriminator. The minimax rivalry between the 2 sub-DNNs enables the model to generate high-quality images, difficult even for humans to distinguish.