Simple approach :
Speedup = CPInon pipeline * Cycle timenon pipeline / [ CPIpipeline * Cycle timepipeline ]
Now by default Cycle timenon pipeline = Cycle timepipeline
Also no of cycles required / instruction in non pipelined system = No of stages( Pipeline )
= 6 here
And by default , CPIideal pipeline = 1
And now we find stall / instruction = Branch frequency * Branch penalty..
= 0.1 * 5
= 0.5
Hence CPIpipeline = 1 + 0.5 = 1.5
So
Speedup = CPInon pipeline * Cycle timenon pipeline / [ CPIpipeline * Cycle timepipeline ]
= 6 / 1.5
= 4
Hence 4 is the correct answer..