Coursehero >> Illinois >> Illinois Tech >> CS 470

Course Hero has millions of student submitted documents similar to the one below including study guides, homework solutions, papers, exam answer keys and textbook solutions.

Myers Donald 1/2 Homework #5 Donald Myers 6.1) a) No, this will not change the gain from pipelining, since pipelining can only move as fast as the slowest operation. b) Yes, this will affect the system if pipelined because the slowest time will now be 25% slower and therefore will cause all operations to be that much longer in the pipeline. 6.2) a) 10^6 instructions X 100 ps to finish an instruction = 1.0 x 10 ^ -4 seconds b) 100 ps / 20 stages = 5 ps per stage. On 20th stage first command completes, every stage after that another command completes. 19 X 5 ps = 95 ps so fill table. 10^6 X 5 ps = 5 X 10 ^ -6 seconds. 95 ps + 5 X10 ^-6 = 5000095 ps. 5000095 ps / 10 ^ -4 = 5% faster c) This will affect both since the overhead will be based on the time taken to initiate and complete every instruction. 6.4) add $3, $4, $2 sub $5, $3, $1 3 will be updated with previous command., need to stall lw $6, 200($3)3 will be updated with add command, with previous nop stall, problem will be solved with forwarding. Add $7, $3, $66 will be a stall since it requires lw to be completed. 6.14) lw is at wb stage therefore, register wb = $ 1052 add instruction is at mem stage, register mem = $24 or instruction is at Ex stage, and therefore forwarded reg 1 = $0 don t care sub instruction at IF stage and therefore don t care again. 6.17) cycle 1 2 3 4 5 add if id ex mem wb sub if id ex mem add if id ex add if id add if registers 6 and 1 are being read for the add $7, $6, $1 instruction and $2 is about to be written to from the first instruction, $2, add $3, $1. 6.21) one nop between every instruction allowing for data forwarding to quickly fill in the second time around. Making the CPI 6 per instruction since the nop will be counted against them. 6.36) a. Branch 1: T-T-T Branch 2: T-T-T-T Branch 3: T-T-T-T-T-T Branch 4: T-T-T-T-T Branch 5: T-T-T-T-T-T-T Accuracy = 60% b. Branch 1:N-N-N Branch 2:N-N-N-N Branch 3:N-N-N-N-N-N Branch 4:N-N-N-N-N Branch 5:N-N-N-N-N-N-N Accuracy = 40% Cs470-051: Fall 2007 Science Illinois Institute of Technology Computer Donald Myers c. 2/2 Homework #5 Branch 1:T-T-T Branch 2:T-N-N-N Branch 3:T-T-N-T-N-T Branch 4:N-T-T-T-N Branch 5:N-T-T-N-T-T-N Accuracy = 44% d. Branch 1:T-T-T Branch 2:T-N-N-N Branch 3:T-T-N-T-N-T Branch 4:T-T-T-T-N Branch 5:T-T-T-N-T-T-N Accuracy = 52 % 6.40) First off if your compiler is good then it will detect the potential data hazard and automatically install a nop after the lw instruction to prevent the program from executing improperly. If that isn t the case with this particular system then it would still seem unlikely that a stall and flush would occur at the same time since the system will have only one command after the branch to deal with. Since it requires to consecutive commands to be executed after a branch in order for stalls to occur it seems unreasonable for said condition to occur. 6.47) lw $2, 0($10) lw $5, 4($10) sub $4, $2, $3 sub $6, $5, $3 sw $4, 0($10) sw $6, 4($10) addi $10, $10, 8 bne $10, $30, Loop Cs470-051: Fall 2007 Science Illinois Institute of Technology Computer

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents: