| US 7,496,736 B2 | ||
| Method of efficient digital processing of multi-dimensional data | ||
| Siamack Haghighi, 15 Cala D'Or, Laguna Niguel, Calif. 92677 (US) | ||
| Filed on Aug. 03, 2005, as Appl. No. 11/196,868. | ||
| Claims priority of provisional application 60/573904, filed on Aug. 27, 2004. | ||
| Prior Publication US 2006/0044316 A1, Mar. 02, 2006 | ||
| Int. Cl. G06F 7/38 (2006.01); G06F 9/00 (2006.01); G06F 9/44 (2006.01); G06F 15/00 (2006.01) | ||
| U.S. Cl. 712—221 | 6 Claims |
| 1. A technique for processing n-dimensional data comprising:
determining a number of dimensions n for data to be processed;
subdividing the data to be processed into elements;
grouping data elements into a plurality of n-dimensional blocks;
identifying a plurality of reference blocks from among the plurality of n-dimensional blocks;
identifying a plurality of search regions, each one corresponding to one of reference blocks;
loading data from a first one of the plurality of search regions corresponding to a first one of the plurality of reference
blocks into a memory array;
processing the data in the memory array with respect to the first one of the plurality of reference blocks;
saving processing results associated with the first one of the plurality of reference blocks;
loading data into the memory array from a portion of a second one of the plurality of search regions corresponding to a second
one of the plurality of reference blocks, wherein the portion does not overlap with the first one of the plurality of search
regions;
processing the data in the memory array with respect to the second one of the plurality of reference blocks;
saving the processing results associated with the second one of the plurality of reference blocks; and
repeating the loading, processing and saving steps until each one of the plurality of reference blocks is processed.
|