Weekly talk 20110126 matlab profiler
Download as PPTX, PDF0 likes286 views
This document discusses using the Matlab Profiler GUI tool and provides 4 examples of ways to speed up Matlab code: 1) Avoiding loops by vectorizing operations, 2) Avoiding growing variables which require copying to new memory locations, 3) Reducing function call overhead by passing arguments rather than defining new functions, and 4) Using sparse matrices instead of full matrices when possible to reduce memory usage and computation time. The examples demonstrate profiling code with the Matlab Profiler and provide specific cases of optimizing code for improved speed and efficiency.
Weekly talk 20110126 matlab profiler
- 1. Matlab Profiler Basic Stanley 2011/1/25 Weekly Talk
- 2. Outline • Matlab Profiler – GUI Profile tool in matlab • Speed Up – Example 1: Avoid Loop – Example 2: Avoid Growing Variable – Example 3: Avoid Function Overhead – Example 4: Adopt Sparse Matrix
- 5. Example 1.1: Avoid Loop M-Lint results No M-Lint messages.
- 6. Example 1.1: Avoid Loop
- 7. Example 1.2: Avoid Loop
- 8. Example 1.2: Avoid Loop
- 9. Example 2: Avoid Growing Variable
- 10. Example 2: Avoid Growing Variable
- 11. Growing Variable (M-Lint resuls in Matlab R2008a)
- 12. Example 3: Avoid Function Overhead
- 13. Example 3: Avoid Function Overhead
- 14. Example 4: Adopt Sparse Matrix
- 15. Example 4: Adopt Sparse Matrix