loop skewing


here's a picture I drew this morning just now while trying to understand loop skewing from a really obscure description. So obscure it's not even really a description, but an example that's supposed to be exactly irregular enough to exemplify every contingency that might arise in the skewing of an everyday 2d array. teaching problems have to have this quality in a way, but it's funny if you try putting everything to be taught into exactly one problem. so funny that I thought I'd like to learn loop skewing that way- seems a certain kind of efficient. But, I haven't understood it yet so the drawing is of the pre-skew traversal of the array in a usual way, showing dependencies of later assignments on prior ones. Then I got a bit frustrated and crumpled up my drawing. hopefully I'll be posting a not frustrated drawing soon-- maybe after pancakes

No comments: