Amdahl's law is suitable for applications where response time is critical. On the other hand, there are a lot of applications which need that accuracy of the resultant output should be high. In the current situation, the computing power has increased substantially due to increase in number of processors attached to a parallel computer. Therefore it is possible to increase the size of the problem i.e., the workload can be enlarged. How does this operate? Gustafson's Law relaxed the restriction of fixed size of difficulty and aimed at using the notion of constant implementation time in order to overcome the sequential bottleneck encountered in Amdahl's Law. This situation which does not suppose a fixed workload is analyzed by Gustafson. Therefore, Gustafson's Law suppose that the workload may increase substantially, with the number of processors but the whole implementation time should remain the similar as highlighted in Figures shown below.
Number of Processors
Parallel Load Increases for Gustafson's Law
According to Gustafson's Law, if the amount of parallel operations for a problem increases sufficiently, then the sequential operations will no longer be a bottleneck.