Exploiting Inter-Warp Heterogeneity to Improve GPGPU Performance

Keyphrases

• GPGPU 100%
• Caching 100%
• Memory Divergence 100%
• Divergence Correction 42%
• Divergence 28%
• Stall 28%
• Queuing Delay 28%
• Shared Cache 28%
• Memory Effect 14%
• Impact Performance 14%
• GPGPU Computing 14%
• Memory-level Parallelism 14%
• Management Techniques 14%
• High Energy Efficiency 14%
• Management Mechanism 14%
• Cache Bypassing 14%
• Cache Management 14%
• Memory Cache 14%
• Long Latency 14%
• Latency Tolerance 14%
• Negative Performance 14%
• Hit Rate 14%
• Memory Controller 14%
• Lockstep 14%
• GPU Memory Management 14%
• Warp Divergence 14%
• Insertion Policy 14%
• Heterogeneous Memory 14%

Computer Science

• Graphics Processing Unit 100%
• Queueing Delay 100%
• Energy Efficiency 50%
• Performance Impact 50%
• Level Parallelism 50%
• Memory Controller 50%
• Memory Instruction 50%
• Hit Rate 50%
• Warp Divergence 50%