MaxEVA: Maximizing the Efficiency of Matrix Multiplication on Versal AI Engine

Keyphrases

• Matrix multiplication 100%
• AI Engine 100%
• Versal 100%
• FP32 42%
• Energy Efficiency 28%
• Deep Learning Workloads 28%
• High Energy Efficiency 14%
• Processor Core 14%
• Memory Requirements 14%
• 32-bit 14%
• 8-bit 14%
• Hardware Architecture 14%
• Performance Bottleneck 14%
• Throughput Gain 14%
• Computational Requirements 14%
• Programmable Processor 14%
• Performance Efficiency 14%
• Xilinx 14%
• Floating-point 14%
• Multiple Angles 14%
• Engine Architecture 14%
• Knowledge Gaps 14%
• Vector-based Algorithm 14%

Computer Science

• Matrix Multiplication 100%
• Artificial Intelligence 100%
• Energy Efficiency 42%
• Deep Learning Method 28%
• Processor Core 14%
• Memory Requirement 14%
• Hardware Architecture 14%
• Performance Bottleneck 14%
• Floating Point 14%