You have written a program that currently takes 1 hour (60 minutes) to run on a high-end desktop computer: reading the data from the input file takes 10 minutes, the computation takes 45 minutes, and writing the output file takes 5 minutes. On a large enough parallel supercomputer, you can speed-up the computation a lot... but not the file I/O. According to Amdahl's law, what is the maximum speedup you could hope to get by using a parallel supercomputer?