Accelerating Neural Network Inference with Processing-in-DRAM: From the Edge to the Cloud

Keyphrases

• Neural Network 100%
• Neural Network Inference 100%
• Processing-in-DRAM 100%
• Neural Network Model 50%
• Performance Efficiency 50%
• Energy Efficiency 50%
• Memory Bound 50%
• Mensa 50%
• Computational Resources 25%
• Google 25%
• Three-state 25%
• Memory Resource 25%
• CPU-GPU 25%
• 3D Stacking 25%
• Viable Solutions 25%
• Energy Efficiency Improvement 25%
• Network Energy 25%
• Edge Devices 25%
• Energy Throughput 25%
• Memory Array 25%
• Binary Neural Network 25%
• Neural Network Performance 25%
• Bit-serial Arithmetic 25%
• Edge TPU 25%
• Memory Oversubscription 25%

Computer Science

• Neural Network 100%
• Network Inference 100%
• Energy Efficiency 42%
• Neural Network Model 28%
• Memory Bound 28%
• Network Performance 14%
• Memory Resource 14%
• Oversubscription 14%
• Memory Array 14%