Consider the following data:

machine $M1$: clock rate $=800MHz$

machine $M2$: clock rate $=700MHz$

running program$-1onM1$ takes $15sec,onM2$ takes $20sec$

running program$-2onM1$ takes $10sec,onM2$ takes $8sec$

For program$-1,$ assume that the number of instructions executed on M$1$ and M$2$ are

$600$ million and $500$ million, respectively.

Find the clock cycles per instruction $($CPI$)$ for program$-1$ on both machines.

Assume that the CPI for program$-2$ on each machine described in #$1$ is the same as

the CPI for program$-1$ on each machine found in #$1.$

Find the instruction count $($IC$)$ for program$-2$ running on each machine using the

execution times shown in #$1.$

Consider two different implementations, machine M$1$ and machine M$2,$ of the same

instruction set architecture as described below.

Note that there are four types of instructions $($A$,$ B$,$ C and D$)$ in the instruction set.

$M1$: clock rate $=800MHz$; $CP{I}_A=1,CP{I}_B=2,CP{I}_C=3,CP{I}_D=4$

$M2$: clock rate $=700MHz$; $CP{I}_A=2,CP{I}_B=3,CP{I}_C=4,CP{I}_D=2$

Assume that the 'peak performance' is defined as the fastest rate that a machine can

execute an instruction sequence chosen to maximize that rate. What are the peak

performances of M$1$ and M$2$ expressed as instructions per second? Answer with MIPS.

If the number of instructions executed in a certain program is divided equally among

the four types of instructions shown in #$3,$ which machine is faster? Please show how

much times faster than the other machine.

Consider the CPI values from #$3$ and the instruction type distribution from #$4.$

At what clock rate would M$1$ have the same performance as the $700$ MHz version of M$2?$

Consider the following data for computer C $1$ :

C $1$ : clock rate $=2\mathrm{.}2GHz,$ running a benchmark program on C $1$ takes $20$ sec $.$

A design team is developing a new computer C $2$ aiming the double performance by

increasing the clock rate with the cost of increased clock cycles with the factor of $1\mathrm{.}4.$

What would be the clock rate of computer C$2?$

Amdahl's law

Suppose that we enhance a computer system to make all floating$-$point instructions run

faster than the original version. Assume that a benchmark program consists of floating

point instructions $(40\%)$ and other instructions $(60\%).$

To achieve the speedup of $1\mathrm{.}6$ for running this benchmark program, what should be the

speedup of the enhanced mode $($floating point part$)?$

Submission: Please write answers on blank papers and submit a $.$pdf version $($single file$).$