September 05, 2006
Inside the VSIPL++ APIConclusion
The functionality I discuss in this article is present in the VSIPL++ 1.0 API. The VSIPL Forum is already discussing functionality to include in future versions of VSIPL++. Users can expect that future versions of the API will contain substantially more image-processing functionality.
VSIPL++ is also emerging as a research vehicle for high-performance computing. A group at Northeastern University (http://www.ece.neu.edu/groups/rcl/projects/vsipl/vsipl.html) has been working on integrating Field-Programmable Gate Arrays (FPGAs) with VSIPL++. (This work leverages VSIPL++'s user-defined storage mechanism, which can also be used to implement sparse arrays, or other custom storage formats.)
Because VSIPL++ was originally designed for use on embedded systems, Sourcery VSIPL++ does not currently run on Windows. However, we plan on adding Windows support soon. We also plan to extend the profiling and debugging capabilities of our optimized implementation, add more optimizations, and provide higher level computational algorithms not required by the API specification.
DDJ
Previous Page |
1 Inside the VSIPL++ API
|
2 Computation
|
3 Implementing in VSIPL++
|
4 Implementing in VSIPL++, Part II
|
5 Performance
|
6 Parallel Computation
|
7 Conclusion
|
 |
|
|||||||||||||||||||||||||||||
|
|
|
