Accelerating Distributed-memory Autotuning via Statistical Analysis of Execution Paths

Keyphrases

• Statistical Analysis 100%
• Prediction Accuracy 100%
• Distributed Storage 100%
• Statistical Profile 100%
• Execution Path 100%
• Predictive Models 50%
• Confidence Interval 50%
• Path Analysis 50%
• Recurring 50%
• Performance Prediction 50%
• Execution Time 50%
• Numerical Linear Algebra 50%
• Shared Memory Architecture 50%
• Profiling Tools 50%
• GPU Architecture 50%
• Invocation 50%
• Execution Method 50%
• Automatic Tuning 50%
• Online Execution 50%
• QR Factorization 50%
• Cholesky Factorization 50%
• Algorithm Configuration 50%
• Selective Kernel 50%

Computer Science

• Distributed Memory 100%
• Execution Path 100%
• Autotuning 100%
• Kernel Execution 66%
• Statistical Profile 66%
• Prediction Accuracy 66%
• Execution Time 33%
• Performance Tuning 33%
• Predictive Model 33%
• Profiling Tool 33%
• Critical Path 33%
• Graphics Processing Unit 33%
• Shared Memory 33%