Tutorial Sheet ODD SEM 2023

Computer Organization and Architecture (15B11CI313)

5th Sem CSE

Jaypee Institute of Information Technology, Noida

Instruction for Tutorials

1. Tutorial has to be submitted (Turnin) within three days of tutorial

class conducted.
2. Graded tutorial sheet will be returned within week of conduct of
tutorial class by the concerned teacher.
3. All tutorial submission has to be handwritten with your name,
enrollment and batch compulsory on it.
4. End of the semester COA project will be evaluated in tutorial class
5. Attendances are compulsory for all tutorial classes

Tutorial -2: Computer Performance Measurement

Tutorial-2 is based on the Module-2 ( Performance Measures For Computer System) which
introduces various measuring metrics to measure computer performance. The references for
the this tutorial is the from the textbook and online resources

i) Computer Architecture – A Quantitative Approach, John L. Hennessy and David

[Link], [Link], Morgan Kaufmann, Elsevier, 2011.

ii) Computer Organization and Design – The Hardware / Software Interface, David A.
Patterson and John L. Hennessy, [Link], Morgan Kaufmann, Elsevier, 2009.

iii) Internet reference for general questions

iv) Online resource Summarizing Performance, Amdahl's law and Benchmarks – Computer
Architecture
v) Online resource COMP 303 Computer Architecture Lecture 5

1) Discuss the relative importance of the various performance metrics discussed in Lecture
Class.
2) Some people argue that in the future, processors will become essentially free as the cost
of computers become dominated by the cost of storage and communication networks.
Discuss how this situation may affect algorithm design and performance analysis.
3) A benchmark program is run on a 200 MHz processor. The The executed program consists of
2 million instruction executions, with the following instruction mix and clock cycle
count:

Instruction Type Instruction Count (%) Cycles per Instruction

Arithmetic and logic 38 1
Load and store 15 3
Branch 42 4
Control transfer 5 2

Determine the effective CPI, MIPS rate, and execution time for this program.

4) Based on your response to the following questions you have to select a faster machine
between machine A and machine B which are running different compilers and their
characteristics are as follows.

MACHINE A – 2 GHz
clock rate
Instruction Class CPI for the class Instruction Count (in billions)
A 4 9
B 1 15
C 5 12
D 2 24

MACHINE B – 2.5 GHz

clock rate
Instruction Class CPI for the class Instruction Count (in billions)
A 1 16
B 3 24
C 1 12
D 5 28

a. What is the average CPI of machine A and B?

b. On which machine is the program faster with respect to
i. Execution time

th
 ii. MIPS rating

5) M1 and M2 are two machines whose performances have to be compared based on the
following measurements.

Program Time on M1 (sec) Time on M2 (sec)

A 30.0 20.0
B 40.0 80.0

Program Instructions executed on M1 Instructions executed on M2

A 7000 Millions 3000 Millions

a. Which machine is faster for each program and by how many times and by what
percentage?
b. Find the instruction execution rate (instructions per second) for each machine for
program A?
c. Find the CPI for program A on both machines (M1: 3GHz, M2: 5GHz)?
d. For a typical workload program A must be executed 100 times each hour and the
remaining time must be used to execute program B. Which machine is faster for this
workload?

6)
a) Overall speedup if we make 80% of a program run 15 times faster
b) Overall speedup if we make 70% of a program run 15% slower
slower.
c) You have a system that contains a special processor for doing floating
floating-point
point operations.
You have determined that 50% of your computations can use the floating-point
floating point processor.
The speedup of the floating pointing
pointing-point
point processor is 20. Calculate overall
over speedup
achieved by using the floating-point
point processor.

7) A program runs in a thousand seconds on a particular machine, with 85% of the time spent
doing multiply/divide operations. We want to redesign the machine to provide it with faster
multiply/divide hardware.
a. How much faster should the multiplier/divider become for the program to run three
times as fast?
b. What if we want the program to run four times as fast?

8. Calculate the effective CPI for an architecture based on the following measurements of
average CPI for various types of instructions:
Instruction type %age of Clock cycles
occurrence
ALU instructions 45% 1.0
Load-stores 38% 1.4
Conditional 15% Taken 2.0

th
 Branches Not 1.5
taken
Jumps 2% 1.2
Assume that 60% of the conditional branches are taken.

9. A benchmark program runs for 100 seconds. We want to improve the speed of the
benchmark by a factor of 4. To achieve this, we enhance the floating
floating-point
point hardware to make
floating point instructions run 10 times faster. How much of the initial execution time would
floating-point
point instructions have to account for, to get the above speedup of 4?

10. You have a system that contains a special processor for doing floating
floating-point
point operations.
You have determined that 60% of your computations can use the floating-pointpoint processor.
When a program uses the floating
floating-point
point processor, the speedup of the floating-point
floating
processor is 40% faster than when it doesn’t use it.

a) Overall speedup by using the floating

floating-point processor.
b) In order to improve the speedup you are considering two options:
• Option 1: Modifying the compiler so that 70% of the computations can use
the floating point processor. Cost of this option is $50K.

• Option 2: Modifying the floating point processor. The speedup of the floating point
processor is 100% faster than when it doesn’t use it. Assume in this case that 50% of the
computations can use the floating point processor. Cost of this option is $60K.

Which option would you recommend? Justify your answer quantitatively.

th