Adv. Computer Architecture CS701 - Jan 011

2
CS 70 1 USN M S 0 M S RAMAIAH INSTITUTE OF TECHNOLOGY (AUTONOMOUS INSTITUTE, AFFILIATED TO VTU) BANGALORE - 560 054 SEMESTER END EXAMINATIONS - JANUARY 2011 Course & Branch : B.E. - Computer Science & Engineering Semester: VII Subject : Advanced Computer Architecture Max. Marks: 100 Subject Code : CS701 Duration: 3 Hrs Instructions to the Candidates: • Answer one full question from each unit • Draw neat figures wherever necessary 1 a) b) C) 2. a) experi UNIT - I What are the important classes of computers? List their important features and differentiate them on the basis of system design issues. What is ISA? Illustrate the seven dimensions of ISA. Consider the execution of a bench mark code with 200,000 instructions on a 40MHz processor. The program consists of four major instruction types. The instruction mix and CPI needed are given below on the result of a trace Instruction Type CPI Instruction mix Arithmetic 1 60% Load store 2 18% Branch 4 12% F.P. 8 10% Calculate the average CPI and the CPU time for the above results. If the CPI of FP operations are reduced to 6, what will be the effect? Calculate the speed up. - b) State Amdahl's law. Suppose that we want to enhance the processor used (06) for web applications. The new processor is 10 times faster on computation in the web application than the original processor. Original processor is busy with computation 40% of time, waiting for I/O 60% of time, what is the overall speed up gained by the enhancement? c) Illustrate how every instruction in RISC subset can be implemented using a (08) pipeline. Briefly discuss the three classes of hazards. (06) (08) (06) What are the different changes in technology that have brought rapid (06) changes in parallel computers? Discuss bandwidth versus latency issue. UNIT - II 3. a) What are the different types of dependencies? Explain with examples. (06) b) How to overcome data hazards using dynamic scheduling? (08) Illustrate the information tables using Tomasulo algorithm for the given code c) With a neat diagram, explain the Pentium 4 micro architecture. sequence when the first two load instructions are completed i) L.D F6, 16(R2) ii) L.D F2, 32(R3) iii) MUL D FO, F2, F4 iv) SUB D F8, F2, F6 (06) 4. a) What is a data hazard? How are they classified? (05) b) Suppose we have a VLIW that could issue two memory references two FP (10) operations and one integer operation or branch in every clock cycle. Show an unrolled version of the loop x[i]=x[i]+S for the given MIPS code. Unroll as many times to eliminate all stalls, ignore delayed branch. Draw relevant figures. Page 1 of 2

description

d

Transcript of Adv. Computer Architecture CS701 - Jan 011

Page 1: Adv. Computer Architecture CS701 - Jan 011

CS70 1USN M S 0

M S RAMAIAH INSTITUTE OF TECHNOLOGY(AUTONOMOUS INSTITUTE, AFFILIATED TO VTU)

BANGALORE - 560 054SEMESTER END EXAMINATIONS - JANUARY 2011

Course & Branch : B.E. - Computer Science & Engineering Semester: VII

Subject : Advanced Computer Architecture Max. Marks: 100

Subject Code : CS701 Duration: 3 Hrs

Instructions to the Candidates:• Answer one full question from each unit• Draw neat figures wherever necessary

1 a)

b)C)

2. a)

experi

UNIT - IWhat are the important classes of computers? List their important featuresand differentiate them on the basis of system design issues.What is ISA? Illustrate the seven dimensions of ISA.Consider the execution of a bench mark code with 200,000 instructions on a40MHz processor. The program consists of four major instruction types.The instruction mix and CPI needed are given below on the result of a trace

Instruction Type CPI Instruction mixArithmetic 1 60%Load store 2 18%

Branch 4 12%F.P. 8 10%

Calculate the average CPI and the CPU time for the above results. If the CPIof FP operations are reduced to 6, what will be the effect? Calculate thespeed up. -

b) State Amdahl's law. Suppose that we want to enhance the processor used (06)for web applications. The new processor is 10 times faster on computationin the web application than the original processor. Original processor is busywith computation 40% of time, waiting for I/O 60% of time, what is theoverall speed up gained by the enhancement?

c) Illustrate how every instruction in RISC subset can be implemented using a (08)pipeline. Briefly discuss the three classes of hazards.

(06)

(08)(06)

What are the different changes in technology that have brought rapid (06)changes in parallel computers? Discuss bandwidth versus latency issue.

UNIT - II3. a) What are the different types of dependencies? Explain with examples. (06)

b) How to overcome data hazards using dynamic scheduling? (08)Illustrate the information tables using Tomasulo algorithm for the given code

c) With a neat diagram, explain the Pentium 4 micro architecture.

sequence when the first two load instructions are completedi) L.D F6, 16(R2) ii) L.D F2, 32(R3)iii) MUL D FO, F2, F4 iv) SUB D F8, F2, F6

(06)

4. a) What is a data hazard? How are they classified? (05)b) Suppose we have a VLIW that could issue two memory references two FP (10)

operations and one integer operation or branch in every clock cycle. Showan unrolled version of the loop x[i]=x[i]+S for the given MIPS code. Unrollas many times to eliminate all stalls, ignore delayed branch. Draw relevantfigures.

Page 1 of 2

Page 2: Adv. Computer Architecture CS701 - Jan 011

CS70ic) List any five important characteristics of the recent Pentium implementation (05)

in 90nm technology.

UNIT - III5. a) What is cache coherence? When is the memory system is said to be (04)

coherent?b) Discuss the protocols used for cache coherence. Show the different states (12)

with a' state diagram for a write invalidate cache coherence protocol for awrite-back cache.

c) What sources influence the coherence misses? Discuss. (04)

6. a) With a neat state transition diagram, explain the Directory Protocol. (08)b) What is a Spin Lock? Consider the following code. (06)

DADDUI R2, R0, #1lock-it: EXCH R2, 0(R1)

BNEZ R2, lock-itExplain how this above code is used as spin lock.

c) Show the organization of T1 processor. List its important features. (06)

UNIT - IV7. a) How a small and simple cache help the cache performance? The hit time of (06)

a two way set associative cache is 1.1 times faster than four way setassociative cache of the same size. The miss rate falls from 0.049 to 0.044to 8KB cache. Hit is 1 clock cycle and miss penalty is 10 clock cycles to theL2 cache for two way cache and L2 cache does not miss. Which has thefaster average memory access time?

b) How compiler optimization can reduce miss rate? Consider the given loop (08)interchange in such a way that it reduces miss rate.for (j=0; j<100; j=j++)

for (i=0; i<5000; i=i++)x[i][j]=2*x[i][j];

c) How the computer architecture facilitates sharing of hardware resources by (06)all processes? What is TLB?

8, a) How a critical word first and early restart reduces the miss penalty in a (03)cache?

b) Write a brief note on the following (09)i) Protection and ISAii) Speculative execution and memory systemiii) I/O and consistency of cached data

c) What are the requirements of a virtual machine monitor? (04)d) List the important components of memory of a AMD opteron processor. (04)

UNIT - V9. a) What is RAID? Summarize the fire standard RAID levels. (07)

b) "A mistake by a human operator is a fault. The resulting altered data is an (06)error. It is latest until activated and leads to failure". Clarify the relationbetween faults, errors and failures.

c) Show the typical response time of a I/O system. How does the architect (07)balance the demands?

10. a) Why power is an important issue in a disc storage system? (06)b) What are bench marks? What are the typical characteristics of a TPC bench (06)

mark?c) Discuss the EPIC architecture in brief. (08)

Page 2 of 2