Scribd
Upload
28 views

Tuto 8

The document contains two problems. The first problem asks to calculate the total time, speedup factor, efficiency, and instruction throughput for a 15-stage pipeline CPU executing 75 instructions. The second problem asks to write efficient ARM assembly code to compute a value z based on values of x, y, and conditions on x.

Uploaded by

rosestrikes
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
28 views

Tuto 8

The document contains two problems. The first problem asks to calculate the total time, speedup factor, efficiency, and instruction throughput for a 15-stage pipeline CPU executing 75 instructions. The second problem asks to write efficient ARM assembly code to compute a value z based on values of x, y, and conditions on x.

Uploaded by

rosestrikes
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Tutorial 8

[1]. Consider 15-stage pipeline CPU having delay of 3ns per stage. (a) Calculate total time to execute 75
instructions by this pipeline CPU. Calculate Speed up factor, Efficiency and Instruction throughput
(inst./ns and inst./clock) neglect latch delay.

[2]. Write an efficient ARM assembly code to compute following where x,y and z are signed numbers.
Assume that x and y are available in registers R0 and R1 and z is to be stored in R2.
if (x < 11 ) z = x + (y/64);
else if (x= =11) z = z – (y*4);
else z = (32* y) – x ;

Ans 1:
No. of stages k=15
Number of instructions n=75
Pipe line clock period tc=3 ns
(a) It will require (15+74) clock cycle to execute 75 instructions in pipeline CPU
Total time in pipeline CPU is Tk= 89 tc => Tk= 267 ns
(b) Speed up factor Sk= Total time in Non-pipeline / Total time in pipeline
= n k tc / [k+(n-1) tc] = 3375/267 = 12.64
- Efficiency Ek = Sk /k => Ek = 0.84
- Instruction throughput of pipeline Hk = n/ Tk => Hk=0.28 inst/ns
Hk= 0.28 inst./ns = 0.84 inst./3ns = 0.84 inst./clock

Ans 2:

CMP R0, #11


ADDLT R2, R0, R1 LSR #6 ; z = x+ y/64
SUBEQ R2, R2, R1 LSL #2 ;z= z- (y*4)
RSBGT R2, R0, R1 LSL #5 ;z=32y-x

You might also like