Abstract | ||
---|---|---|
The use of highly optimized inner kernels is of paramount importance for obtaining efficient numerical algorithms. Often, such kernels are created by hand. In this paper, however, we present an alternative way to produce efficient matrix multiplication kernels based on a set of simple codes which can be parameterized at compilation time. Using the resulting kernels we have been able to produce high performance sparse and dense linear algebra codes on a variety of platforms. |
Year | DOI | Venue |
---|---|---|
2006 | 10.1007/11751649_84 | ICCSA (5) |
Keywords | Field | DocType |
hand-coded inner kernel,high performance sparse,simple code,inner kernel,efficient numerical algorithm,compilation time,paramount importance,efficient alternative,dense linear algebra code,efficient matrix multiplication,compiler-optimized kernel,matrix multiplication,compiler optimization | Linear algebra,Parameterized complexity,Matrix calculus,Computer science,Algorithm,Compiler,Linear code,Kernel method,Matrix multiplication | Conference |
Volume | ISSN | ISBN |
3984 | 0302-9743 | 3-540-34079-3 |
Citations | PageRank | References |
4 | 0.43 | 18 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
José R. Herrero | 1 | 94 | 16.90 |
Juan J. Navarro | 2 | 323 | 42.90 |