Sourcery VSIPL++
Next
Sourcery VSIPL++
User's Guide
Version 2.2-9
CodeSourcery, Inc.
Version 2.2-9
Copyright © 2005-2009 CodeSourcery, Inc.
All rights reserved.
Table of Contents
I. Working with Sourcery VSIPL++
1. API overview
1.1. Views
1.2. Blocks
1.3. Matlab IO
2. Direct Data Access
2.1. Introduction
2.2. Basic usage
2.3. Non-dense blocks
2.4. Requesting a specific data layout
3. Using the Dispatch Framework
3.1. Introduction
3.2. Compile-time dispatch
3.3. Runtime dispatch
4. Custom Expression Evaluation
4.1. The problem
4.2. Expression templates
4.3. Expression templates in Sourcery VSIPL++
4.4. Creating custom expressions
4.5. Creating custom evaluators
5. Profiling
5.1. Enabling Profiling
5.2. Using the Profiler
5.3. Profiler Output
6. Benchmarking
6.1. Overview
6.2. Obtaining the Benchmarks
6.3. Running Performance Tests
6.4. Benchmark Output
Glossary
II. Example Application
7. Fast Convolution
7.1. Fast Convolution
7.2. Serial Optimization: Temporal Locality
7.3. Performing I/O with User-Specified Storage
7.4. Performing I/O with External Data Access
8. Parallel Fast Convolution
8.1. Parallel Fast Convolution
8.2. Improving Parallel Temporal Locality
8.3. Performing I/O
A. Benchmark Options
B. Benchmark Usage
B.1. CELL Benchmark Usage
B.2. conv
B.3. conv2d
B.4. copy
B.5. corr
B.6. CUDA Benchmark Usage
B.7. CVSIP Benchmark Usage
B.8. dist_vmul
B.9. dot
B.10. fastconv
B.11. fft
B.12. fftm
B.13. FFTW3 Benchmark Usage
B.14. fir
B.15. HPEC_KERNEL Benchmark Usage
B.16. IPP Benchmark Usage
B.17. LAPACK Benchmark Usage
B.18. maxval
B.19. mcopy
B.20. meansqval
B.21. meanval
B.22. memwrite
B.23. memwrite_simd
B.24. MPI Benchmark Usage
B.25. prod
B.26. prod_var
B.27. pwarp
B.28. SAL Benchmark Usage
B.29. sfilt
B.30. sort
B.31. stream
B.32. sumsqval
B.33. sumval
B.34. sumval_simd
B.35. svmul
B.36. task_switch
B.37. vdiv
B.38. vma
B.39. vmagsq
B.40. vmmul
B.41. vmul
B.42. vmul_c
B.43. vmul_par