Barry Smith <[email protected]> writes:
>    All true. Switching the CG “norm” does not effect the algorithm (in
>    exact arithmetic) it only affects the norm that is printed and when
>    the algorithm stops. The same minimization principles hold
>    independent of the “norm” used.

Yup, and the natural norm avoids the need for an extra reduction.  If
you're using classic (non-pipelined and not single_reduction) CG in a
latency-sensitive environment, then you might benefit from using the
natural norm.

Attachment: pgpw4gmk5sAPa.pgp
Description: PGP signature

Reply via email to