Optimizing memory-bound SYMV kernel on GPU hardware accelerators

Computer Science

• Hardware Accelerator 100%
• Memory Bound 100%
• Graphics Processing Unit 100%
• Precision Arithmetic 33%
• Single Precision 33%
• Multicore 33%
• Underlying Hardware 33%
• Preprocessing Step 33%
• Memory Access 33%
• Concurrent Execution 33%
• Execution Context 33%
• Profiling Tool 33%
• High-Level Programming Language 33%
• Asymptotic Result 33%
• Symmetric Matrix 33%
• Double Precision 33%
• Matrix Vector Product 33%
• Natural Sciences Computing 33%