This preview shows page 1. Sign up to view the full content.
Unformatted text preview: , devoting around 5 to 20 milliseconds to each task at a time. At this rate, the user perceives the tasks as being performed simultaneously, since a human cannot discern time durations shorter than around 100 ms. Within that time the processor can execute millions of instructions. Figure 9.1 plots the durations of different event types on a logarithmic scale, with microscopic events having durations measured in nanoseconds and macroscopic events having durations measured in milliseconds. The macroscopic events are managed by OS routines that require around 5,000 to 200,000 clock cycles. These time ranges are measured in microseconds (abbreviated s, where is the Greek letter “mu”). Although that may sound like a lot of computation, it is so much faster than the macroscopic events being processed that these routines place only a small load on the processor. Practice Problem 9.1:
When a user is editing ﬁles with a real-time editor such as EMACS, every keystroke generates an interrupt signal. The operating system must then schedule the editor process to take the appropriate action for this keystroke. Suppose we had a system with a 1 GHz clock, and we had 100 users running EMACS typing at a rate of 100 words per minute. Assume an average of 6 characters per word. Assume also t...
View Full Document