Next:
Accumulate into a scalar
Up:
Motivation: an example
Previous:
Memory system
Performance
The initial version runs in 118 seconds.
The matrix multiplication takes
steps, each involving two floating-point operations, an add and a multiply, i.e.
.
This loop achieves a computation rate of 250/118=2.1 MFLOPs.
This machine is advertised as having a peak performance of more than 10 MFLOPs.
How are we going to get value for money?
Paul H J Kelly Thu Dec 4 18:15:31 GMT 1997