ARTEMIS: A Mixed Analog-Stochastic In-DRAM Accelerator for Transformer Neural Networks

Computer Science

• In-Memory Processing 100%
• And-States 33%
• Natural Language Processing 33%
• Recurrent Neural Network 33%
• Convolutional Neural Network 33%
• Parallelism 33%
• Performance Gain 33%
• Execution Time 33%
• Attention (Machine Learning) 33%
• Execution Model 33%
• Massive Amount 33%
• Computer Architecture 33%
• Hardware Accelerator 33%
• Memory Footprint 33%
• Memory Bandwidth 33%
• Graphics Processing Unit 33%
• Computer Vision 33%