Siracusa: A 16 nm Heterogenous RISC-V SoC for Extended Reality With At-MRAM Neural Engine

Keyphrases

• Non-volatile Memory 100%
• RISC-V 100%
• Magnetic Random Access Memory 100%
• Extended Reality 100%
• Neural Engine 100%
• Siracusa 100%
• Machine Learning 66%
• Energy Efficiency 66%
• Deep Neural Network 66%
• System-on-chip 66%
• System Level 33%
• Signal Processing 33%
• High-throughput 33%
• Signal Control 33%
• Memory-based 33%
• Memory Access 33%
• On chip 33%
• Tightly Coupled 33%
• Machine Learning Performance 33%
• Conventional Signal 33%
• Average Power 33%
• Optical Coherence Tomography Angiography (OCT-A) 33%
• Chip System 33%
• XR Applications 33%
• Area-efficient 33%
• Network Weight 33%
• Heterogeneous SoC 33%
• Peak Energy 33%
• Application Workload 33%
• Memory Integration 33%
• Heterogeneous Applications 33%
• Learning Efficiency 33%
• Eureka 33%
• Power Constrained 33%
• Latency Bound 33%
• Low-power System 33%
• Digital Signal Processing 33%
• Deep Neural Network Inference 33%

Computer Science

• Extended Reality 100%
• Random Access Memory 100%
• System-on-Chip 100%
• Non-Volatile Memory 60%
• Machine Learning 60%
• Learning System 60%
• Deep Neural Network 60%
• Energy Efficiency 40%
• Learning Performance 20%
• Memory Access 20%
• Processing Core 20%
• Network Inference 20%
• Digital Signal Processing 20%
• High Throughput 20%
• Digital Art 20%

Engineering

• Magnetoresistive Random-Access Memory 100%
• System-on-Chip 100%
• Nonvolatile Memory 60%
• Learning System 60%
• Deep Neural Network 60%
• Energy Efficiency 40%
• Energy Conservation 40%
• Energy Engineering 20%
• Power Engineering 20%
• Heterogeneous Sensor 20%
• Power Level 20%
• Memory Access 20%
• Peak Energy 20%
• Digital Signal Processing 20%
• Average Power 20%