| | |
Summary: Optimizing Matrix Multiply using PHiPAC: a Portable,
High-Performance, ANSI C Coding Methodology
Je Bilmes, Krste Asanovi
cy, Chee-Whye Chinz, Jim Demmelx
fbilmes,krste,cheewhye,demmelg@cs.berkeley.edu
CS Division, University of California at Berkeley
Berkeley CA, 94720
International Computer Science Institute
Berkeley CA, 94704
Abstract
Modern microprocessors can achieve high performance
on linear algebra kernels but this currently requires ex-
tensive machine-speci c hand tuning. We have devel-
oped a methodology whereby near-peak performance on
a wide range of systems can be achieved automatically
for such routines. First, by analyzing current machines
and C compilers, we've developed guidelines for writing
Portable, High-Performance, ANSI C PHiPAC, pro-
nounced fee-pack". Second, rather than code by hand,
we produce parameterized code generators. Third, we
write search scripts that nd the best parameters for a
|
