A computer has an average interrupt response time of p; takes q to process an interrupt; and takes s to return from interrupt. If a polled interrupt mechanism is implemented and it takes x to poll a device and y to service a device that interrupted, how many devices can be polled before polling becomes less efficient than interruptdriven I/O?
What will be an ideal response?
The time taken to deal with an interrupt is p + q + s. The time taken to deal with a polled I/O transaction is n.x + y,
where n is the number of devices polled. For the polling time to be less than the interrupt time, then
n.x + y ? p + q + s; that is, n ? ( p + q + s ? y)/x
You might also like to view...
To move to the previous slide in Slide Show view, _____________.
A. click the right mouse button B. press the left-arrow key C. either a or b D. neither a nor b
The Intel Corporation introduced the ________, which was a chip that contained millions of transistors
Fill in the blank(s) with correct word
A _____ is a central storehouse of information about a system’s data.
A. data cube B. data feed C. data mart D. data repository
Each copy of a master workbook you distribute must be designated as shared, and the ____ feature on the Advanced tab of the Share Workbook dialog box must be activated.
A. Compare History B. Merge History C. Change History D. Review History