Suppose we have a benchmark that executes in 100 seconds of elapsed time, of which 90 seconds is CPU time and the rest is 1/0 time. Suppose the number of processors doubles every two years, but the processors remain the same speed, and I/O time doesn't improve. How much faster will our program run at the end of six years?

Suppose we have a benchmark that executes in 100 seconds of elapsed time, of which 90 seconds is CPU time and the rest is 1/0 time. Suppose the number of processors doubles every two years, but the processors remain the same speed, and I/O time doesn't improve. How much faster will our program run at the end of six years?

Database System Concepts

7th Edition

ISBN:9780078022159

Author:Abraham Silberschatz Professor, Henry F. Korth, S. Sudarshan

Publisher:Abraham Silberschatz Professor, Henry F. Korth, S. Sudarshan

Chapter1: Introduction

Section: Chapter Questions

Problem 1PE

See similar textbooks

Similar questions

Refer to the code below determine whether this code is correct, and if not, correct any mistakes. Find the speedups and efficiencies of the parallel odd-even transposition. Does the program obtain linear speedups? Is it scalable? Is it strongly scalable? Is it weakly scalable? After correcting the mistakes, execute the code in CUDA or OpenMP and measure any performance improvement in terms of speedups and scalability. Process Pi evenprocess =(i%2 ==2); evenphase = 1; for(step =0; step < n; step++,evenphase = !evenphase) { if((evenphase && evenprocess) || (!evenphase && !evenprocess) { send (&a, Pi+1); recv (&x, Pi+1); if (x<a) a =x; } else { recv (&x, Pi-1); send (&a, Pi-1); if (x > a) a = x; } }
In the picture attached, the Say that the system is running 3 jobs, A, B, and C, and that all of them are CPU-intensive (i.e., each one does one infinitely long CPU burst). The system begins with A on the CPU at the beginning its time quantum while B and C are in the Ready Queue, in that order. a) Show the execution pattern (as a string of A’s, B’s, C’, and o’s) assuming that the scheduler time quantum is equal to 4 ms. Show the execution for more than 20ms (but less than 30ms). b) In the long run (i.e, assuming jobs don’t ever terminate), what percentage of the CPU time is wasted doing context-switching/scheduling? The answers I came up with is as follows: a) AAAAoBBBoCCCoAAAoBBBoCCCoAAA, because for the first 4ms, A gets to execute it's full time quantum, however, when context-switching to the next job, the switch is included in the time quantum. b) 2.77% spent context switching, divide the number of o's by the total amount of runtime.
Consider a computer running a program that requires 250 seconds, where 70 seconds is spent executing FP instructions, 55 seconds for INT operations, 85 seconds for load/save operations, and 40 seconds for branch instructions. By how much is the time for INT operations reduced if the total time is reduced by 20%?
Now consider the Verilog version of this same simple multicycle implementation: http://aggregate.org/CPE380/multiv.html (Links to an external site.) . Which of the following statements about how that works is false? Group of answer choices The bench module instantiates a processor called PE, and generates the clock input signal to it Symbolic names for things like word size and the implementation of each control signal are given using `define The main memory is a reg array defined in a separate memory module, which is instantiated within the processor module The ALU is specified as using a ripple-carry adder rather than carry lookahead, etc. The control logic is implemented by a case statement that performs the appropriate action(s) for the current STATE
Talk about how this computer's addressable space is determined succinctly using an example.
The Java Virtual Machine (and your computer’s assembly language) support a goto instruction so that the search can be “wired in’’ to machine code, like the program at right (which is exactly equivalent to simulating the DFA for the pattern as in KMPdfa, but likely to be much more efficient). To avoid checking whether the end of the text has been reached each time i is incremented, we assume that the pattern itself is stored at the end of the text as a sentinel, as the last M characters of the text. The goto labels in this code correspond precisely to the dfa[] array. Write a static method that takes a pattern as input and produces as output a straight-line program like this that searches for the pattern. Straight-line substring search for A A B A A A int i = -1;sm: i++;s0: if (txt[i]) != 'A' goto sm;s1: if (txt[i]) != 'A' goto s0;s2: if (txt[i]) != 'B' goto s0;s3: if (txt[i]) != 'A' goto s2;s4: if (txt[i]) != 'A' goto s0;s5: if (txt[i]) != 'A' goto s3; return i-8;
Can a single process get stalled in the middle of execution? Your answers should make sense and adhere to logic.
How can we prevent a CPU from wasting time by trying to run too many programmes at once? How does the absence of certain resources affect the viability of carrying out this strategy? Give an example of the results you may anticipate from them.
How can we prevent CPUs from wasting time multitasking? Without such money, how can we implement this plan? Show a sample result.
Suppose you have a finite state machine that accepts bit strings divisible by 11. The FSM has 11 states R0, R1, ... , R10, each corresponding to the remainder when dividing by 11. Input is fed to the FSM one bit at a time from left to right, the same way we did in our in-class examples. If the machine is currently in state R7, an input of 1 should transition it to which state? Type the subscript only of the new state. Your answer should be an integer between 0 and 10, inclusive.
Consider the following set of processes A, B, C, D with the following CPU burst time and I/O. Find the average waiting time with RR of quantum 10ms and context switch time=2ms. What is the average waiting time? Write the waiting time for processes A,B,C,D.
Suppose that a multiprogrammed system has a load of N processes with individual execution times of t1, t2, ...,tN. Answer the following questions: a) How would it be possible that the time to complete the N processes could be as small as: maximum (t1, t2, ...,tN)? b) How would it be possible that the total execution time, T > t1+ t2+ ...+tN? In other words, what would cause the total execution time to exceed the sum of individual process execution times?