User Account

All Pages

hw7, Data Structure & Algorithms

Assignment Help:
Handout 15
COMP 264: Introduction to Computer Systems (Section 001)
Spring 2013
R. I. Greenberg
Computer Science Department
Loyola University
Water TowerCampus, Lewis Towers 524
820 N. Michigan Ave.
Chicago, Illinois 60611-2147
Assignment #7
Issued 4/12
Due (at class time) 4/22
Best to hand in all solutions through the online submission mechansim of the course web
page. No fancy diagramming is required; you can scan something hand-drawn, or use any
simple online tool of your choosing. Just make sure you end up with a common format
that will be easy to handle, for example .pdf, .jpg, .txt or even .doc or .docx if you must.
Also make sure that your submission name is appropriate to the le type, e.g., 7.pdf or
7.txt, etc. (You also can submit directly on shannon/knuth if you are working there by
copying your le to the directory ~rig/c264hw7sub with lename in the form Email-X.EXT,
where Email is your email address, X is a string of eight or more \random" alphanumeric
characters, and EXT is the le type (i.e. do a command similar to, the following with the
correct replacements for EXT, YOUREMAILADDRESS, and RANDOM:
cp 7.EXT ~rig/c264hw7sub/YOUREMAILADDRESS-RANDOM.EXT
)
HW7-1 (43 points) Suppose we wish to write a procedure that computes the inner
product of two vectors u and v. An abstract version of the function has a CPE of 16{17
with x86-64 and 26{29 with IA32 for integer, single-precision, and double-precision data. By
doing the same sort of transformations we did to transform the abstract program combine1
into the more ecient combine4, we get the following code:
void inner4(vec_ptr u, vec_ptr v, data t *dest) {
long int i;
int length = vec_length(u);
data_t *udata = get_vec_start(u);
data_t *vdata = get_vec_start(v);
data_t sum = (data_t) 0;
for (i = 0; i < length; i++){
sum = sum + udata[i] * vdata[i];
}
*dest = sum;
}
Our measurements show that this function has a CPE of 3.00 for integer and
oating-
point data. For data type float, the x86-64 assembly code for the inner loop is as follows
(udata in %rbx, vdata in %rax, limit in %rcx, i in %rdx, sum in %xmm1):
.L87: ; loop:
movss (%rbx,%rdx,4), %xmm0 ; Get udata[i]
mulss (%rax,%rdx,4), %xmm0 ; Multiply by vdata[i]
addss (%xmm0, %xmm1 ; Add to sum
addq $1, %rdx ; Increment i
cmpq %rcx, %rdx ; Compare i:limit
jl .L87 ; If <, goto loop
Assume that the functional units have the latencies and issue times given in Figure 5.12
(and in the course notes).
A. Diagram how this instruction sequence would be decoded into operations, and show how
the data dependencies between them would create a critical path of operations in the style
of Figures 5.13 (fpb-sequential.ppt) and 5.14 (fpb-
ow.ppt and fpb-
ow-abstract.ppt). (25
points.)
B. For data type float, what lower bound on the CPE is determined by the critical path?
(6 points.)
C. Assuming similar instruction sequences for the integer code as well, what lower bound on
the CPE is determined by the critical path for integer data? (6 points.)
D. Explain how the two
oating-point versions can have CPEs of 3.00 even though the
multiplication operation requires either 4 or 5 cycles. (6 points.)
HW7-2 (27 points) Write a version of the inner product procedure described in the
previous problem that uses four-way loop unrolling. (11 points.)
For x86-64, our measurements of the unrolled version give a CPE of 2.00 for integer data
but still 3.00 for both single and double precision
oating point.
A. Explain why any version of any inner product procedure cannot achieve a CPE less than
2.00. (8 points.)
B. Explain why the performance for
oating-point data did not improve with loop unrolling.
(8 points.)

Related Discussions:- hw7

Maximum numbers of nodes a binary tree of depth d, Maximum numbers of nodes...

Maximum numbers of nodes a binary tree of depth d The maximum numbers of nodes a binary tree of depth d can have is 2 d+1 -1.

Recursive function, The location of a node in a binary search tree is defin...

The location of a node in a binary search tree is defined as a string such as LLRRL, which represents the node that you find by starting at the root, and traversing Left, traverse

Implementation of tree, The most common way to insert nodes to a general tr...

The most common way to insert nodes to a general tree is to first discover the desired parent of the node you desire to insert, and then insert the node to the parent's child list.

Comp. sci algorithms, 1. develop an algorithm which reads two decimal numbe...

1. develop an algorithm which reads two decimal numbers x and y and determines and prints out wether x>y or y>x. the input values, x and y, are whole number > or equal to 0, which

Stack, using a program flowchart design a program to illustrate pop and pus...

using a program flowchart design a program to illustrate pop and push operation

Ruby implementation of the symbol abstract data type, Ruby implementation o...

Ruby implementation of the Symbol ADT Ruby implementation of the Symbol ADT, as mentioned, hinges on making Symbol class instances immutable that corresponds to the relative la

What are stored and derived attributes, What are stored and derived attribu...

What are stored and derived attributes?  Stored attributes: The attributes kept in a data base are called stored attributes.  Derived attributes: The attributes that are

Best case, Best Case: If the list is sorted already then A[i] T (n) = ...

Best Case: If the list is sorted already then A[i] T (n) = c1n + c2 (n -1) + c3(n -1) + c4 (n -1)  = O (n), which indicates that the time complexity is linear. Worst Case:

Search on a heap file, Consider the file " search_2013 ". This is a text fi...

Consider the file " search_2013 ". This is a text file containingsearch key values; each entry is a particular ID (in the schema given above). You are tosimulate searching over a h

Disadvantages of the lifo costing method, The disadvantages or limitations ...

The disadvantages or limitations of the last in first out costing method are: The election of last in first out for income tax purposes is binding for all subsequent yea

Write Your Message!

Captcha
Order Assignment

Featured Services

Order Now

Popular Subjects

Free Assignment Quote

Assured A++ Grade

Get guaranteed satisfaction & time on delivery in every assignment order you paid with us! We ensure premium quality solution document along with free turntin report!

Submit Assignment

Academics

Major Subjects

Majors

Get In Touch

whatsapp: +1-415-670-9521

Phone: +1-415-670-9521

Email: [email protected]

TERMS & POLICIES

HELP & SUPPORT

All rights reserved! Copyrights ©2019-2020 ExpertsMind IT Educational Pvt Ltd