CSI 312 Final, Fall 2010

Question #2 (25 points): a) Discuss the performance enhancements gained due to Pipelining a processor. + peta eagles Sera Qt matrattins Bach whucka vy dhurted fo severol ree The Pipelining erobles penerol insfraatins bo a ele om he Lu. Oe Howe houng be wor € Hl) iifva din be Zrush ond dart gasther. WS % Be ic aaa i in He Some Colon Ue Wes nM db Pecos or Coatio! hreadd Nie all Wed of. Con be pycated Clbey a wamcstion #3 (40 points): bichon Sequence of machine instructions will be executed on a register Load Ra, MLa o Lame =F A htutiroy OP Load Rb, MLb Xe cL pe ae. * 4 a Load Re, MLce oe Clock hate Load Rd, MLd . Mul Ra, Rd Beco - Sub Rb, Ra . This block of 3 arithmetic instructions wee . Add Re, Rb . Repeats 1000 times ) — > Store Ra, MLa an WEN es of Ans Aru cAons =4 430043 = |, Store Rb, MLb , Cae eta (_ Store Rc, MLc here each instruction needs a) In the first experiment, we used bere ro \ the execition only one clock cycle to terminate, anda clock rate 'z. What will be the i ) CHES \ im ib de sequence on this processor? (Show your work) time of the above co q| a : \ Frecut\op mage Ssctnt Bee tayo” = Seat H8 206 |U Hr \vow i ted where most of the instructions ment, a multi-cycle CPU is tes b) In the second experi lock cycles to ter in: if is 200 MHz. What will be the need four c! y' minate, and the clock at 2 ; nt - e =e ee: cle processor? What is the speedup over the single cycle execution ti | processor? (Explain your results) Clea 4 Clock bate = Dee UC Soot K. - Zalit as* e 4) 20 FOO xo° se EyecuWvror Ki we) => J \wo || | ktale