CS267: Lecture 15
Sparse Matrix -Vector Multiply
October 17, 2001
Lecturer: Kathy Yelick
Abstract
We explore the design space for sparse matrix-vector multiplication, including
optimizations for register blocking, cache blocking, exploiting symmetry, using
multiple vectors, and reordering.
2001 Lecture Notes
PowerPoint, Postscript,
PDF
Assignments
Homework 3: Conjugate Gradient due 11/7
Final projects will be due Friday, December 14th. (A poster session
will be scheduled sometime earlier the week of the 10-14th.)
Readings