Dense Dynamic Blocks: Optimizing SpMM for Processors with Vector and Matrix Units Using Machine Learning Techniques

Keyphrases

• Matrix Units 100%
• Sparse Matrices 100%
• Machine Learning Techniques 100%
• Unit Vector 100%
• Dynamic Blockage 100%
• Dense Matrix multiplication 100%
• Matrix multiplication 28%
• Floating-point 21%
• New Matrix 14%
• Dense Matrix 14%
• Hybrid Method 7%
• Single chip 7%
• Unique Characteristics 7%
• Chip Module 7%
• Single Precision 7%
• Double Precision 7%
• Small Matrices 7%
• Matrix Operation 7%
• Irregular Memory Access 7%
• Sparsity Pattern 7%
• Hybrid Strategy 7%
• Access Behavior 7%
• Prediction Mechanism 7%
• Basic Linear Algebra Subprograms 7%
• Cache Optimization 7%
• Matrix Pair 7%

Engineering

• Machine Learning Technique 100%
• Matrix Multiplication 100%
• Dynamic Block 100%
• Floating Point 30%
• Functional Unit 10%
• Single Chip 10%
• Hybrid Approach 10%
• Subprogram 10%
• Matrix Operation 10%
• Linear Algebra 10%
• Memory Access 10%
• Sparsity Pattern 10%
• Rich Environment 10%
• Hybrid Strategy 10%

Computer Science

• Machine Learning Technique 100%
• Matrix Multiplication 100%
• Matrix Multiply 40%
• Floating Point 30%
• Point Throughput 30%
• Functional Unit 10%
• Speed-up 10%
• Hybrid Approach 10%
• Single Precision 10%
• single-chip 10%
• Memory Access 10%
• Matrix Operation 10%
• Double Precision 10%
• Sparsity Pattern 10%
• Hybrid Strategy 10%
• Cache Optimization 10%