

Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Solutions to the problems presented in the mips assembly midterm examination held on october 11, 2006. Answers to questions related to writing c code, identifying raw, war, and waw conflicts, optimizing assembly code with register renaming and loop unrolling, and analyzing the performance of a speculative processor.
Typology: Study notes
1 / 2
This page cannot be seen from the preview
Don't miss anything!


Date: October 11, 2006 Name: Student Code: Instructions: This exam is open book and open notes. Allotted time is 50 minutes. Explicitly state all your assumptions. Don’t just give answers, always try to write down the way you were thinking. Even if you can not find a solution, write down the paths you have tried. Problem 1 (50 points) Consider the following assembly program written for a 32 bit MIPS machine with no delayed branches. li r20, 0 # load immediate li r19, 16000 li r15, 0 start: ld r1, 0(r20) # load into r1 from memory add r15, r15, r addi r20, r20, 32 # add immediate sub r3, r19, r bnez r3, start # jump to start if r3 not equal with zero a) Write a C, C++ or Java program which corresponds to this assembly program b) Identify the RAW, WAR and WAW conflicts in the assembly program. c) Assume that the program is run on a machine where there is a stall of 3 cycles between the writing and the reading of a register^1. Rewrite the assembly program by inserting NOP operations where there will be stalls. d) Unroll the loop in the program 4 times in software (assembly). Transform the program by register renaming such that the number of stalls is minimized. (^1) That is, you need 3 NOPs (or other operations) between ADD R1,R2,R3 and ADD R6, R1,R5.
Problem 2 (50 points) Suppose that a processor with a load/store architecture and no delayed branches executes at a clock rate of 2GHz. Arithmetic and logic instructions require 1 cycle, load and store operations 2 cycles, and conditional branches 3 cycles because of the control hazard involved. The typical applications run on this processor contain a mix of 70% arithmetic and logic instructions, 10% load and store instructions and 20% conditional branches instructions. An engineer proposes a modification in the architecture which introduces speculation. The branch prediction algorithm would be correct 80% of the time. When correct, branches would take 1 cycle, when incorrect, 3 cycles as before. However, the modification requires the reduction of the clock frequency to 1.5GHz. a) What is the average cycles per instruction of the original processor? b) What is the average cycles per instruction of the speculative processor? c) Is the speculative processor faster or slower than the original one? By how much?