Unums and Posits™


Curriculum Vitae

Gustafson's Law

In 1988 Dr. Gustafson wrote the paper Reevaluating Amdahl's Law, which later became referred to as "Gustafson's Law" or the "Gustafson-Barsis Law." Dr. Gustafson summarizes this work:
"...The model is not a contradiction of Amdahl's law as some have stated, but an observation that Amdahl's assumptions don't match the way people use parallel processors. People scale their problems to match the power available, in contrast to Amdahl's assumption that the problem is always the same no matter how capable the computer.
"A firestorm resulted after this paper was published. Defenders of traditional serial thinking had believed Amdahl's model allowed them to dismiss parallel processing as academic foolishness, and many scrambled to reassert the validity of Amdahl's assumptions. Now, few argue that parallel processing isn't viable; this paper marked a turning point in the attitude of the computing industry toward parallel processing."

As mentioned, this publication has generated volumes of discussion and commentary in the ensuing years. Gustafson's Law has also become a standard part of the parallel-processing academic curriculum. Links to some of this discussion and to course material appear on the Web Links page.