CS232 Fall 2006 Practice Final December 10, 2006
1. MIPS:
Part A:
Translate the function func into MIPS. Assume that the function fhas already been translated. You must
observe all calling conventions.
int func(int a, int n) {
if(n == 0)
return a;
return f(func(a, n - 1));
}
Part B:
In a high-level language like C, it is possible for the return-type of a function to be a pointer to another
function. Explain how you would write a MIPS function whose return value is the function func from Part A.
Hint: Assume that the argument $a0 to your function is the address of a large chunk of memory to which
you can write. Your function should eventually return the value $a0.
2. Pipelining and ECC:
This problem asks you to apply the concept of pipelining to networks. A sender encodes the data as packets
(including parity bits for error detection) and transmits them over the network. The receiver inspects
the packets and sends an acknowledgement indicating whether or not an error was detected. Thus, the
transmission process for each packet can be broken into five stages:
Create (C): Create packet (data plus parity bit).
Send (S): Send the packet across the network.
Wait (W): Wait until an acknowledgement is received.
Receive (R): Receive acknowledgement packet.
Decode (D): Decode acknowledgement packet. If necessary, resend the original packet.
These five tasks are handled by independent hardware, and can therefore be pipelined. The latency of each
stage is given below:
Stage Latency (ms)
C 1.0
S 2.5
W 2.0
R 2.0
D 0.7
Part A:
(a) Compute the latency of each packet and maximum throughput (in packets per second) assuming a
5-stage pipeline.
(b) Any packet that gets corrupted must be re-sent. If every packet has a 10% probability of getting
corrupted, what is the expected time to send a packet?
Part B:
(a) Draw a pipeline diagram to indicate what the sender does when the acknowledgement indicates that
an error was detected.
1
