M
C
A
B
N
K
N
M
K
*
NB
Adaptive Approach for Basic Linear
Algebra Subprograms
¨Do
a parameter study of the operation on the target
machine, done once.
¨Only
generated code is on-chip multiply
¨BLAS
operation written in terms of generated
on-chip multiply
¨All
tranpose cases coerced through data copy to 1 case
of on-chip multiply
ãOnly 1 case generated per platform