Project Planning ools hapter 7 Project Planning Lecture 2 Project Planning nd Management ools GN hart (Figure 7.) PM hart (Figure 7.2) P hart (Figure 7.5) GN: eveloped by Henry Gantt PM: ritical Path Method P: Program valuation and eview echnique lide # lide #2 GN hart Projected/actual progress vs. time ime: horizontal axis ctivity: vertical axis dv: learly shows timeline, Projected vs. actual progress Project milestones is: oes not easily show task dependencies/precedence lide # lide #4
lide #5 lide #6 lide #7 lide #8
lide #9 lide # chedule and Milestones chedule and Milestones evelop and Validate Power Models for M core xtend Power Models to trongm emonstrate on trongm power*performance improvements evelop and Validate Power Models for trongm core plus P bus to memory esign custom trongm/xcale P xtend Power Models to Xcale/trongM2 plus second level memory Validate custom trongm/xcale Printed ircuit oard xtend and Validate rimaran for P/ hallenge pplications xtend and Validate rimaran for P/ hallenge pplications emonstrate on trongm/xcale energy*performance improvements emonstrate avings through lgorithm-ased ransformations emonstrate dynamic on-demand adaptability to changing energy constraints based on code swapping emonstrate avings through lgorithm-ased ransformations emonstrate dynamic on-demand adaptability to changing energy constraints based on code swapping ompiler Optimizations Validate ode Generator for M core emonstrate Power/Performance radeoffs using M/trongM models in IMN xtend ode Generator to trongm emonstrate savings possible through use of advanced technology features ontinue work on Machine ependent ompiler Optimizations, update ode Generator ompiler Optimizations Validate ode Generator for trongm core emonstrate Power/Performance radeoffs using trongm/xcale models in IMN xtend ode Generator to Xcale/ trongm2 emonstrate savings possible through use of advanced technology features ontinue work on Machine ependent ompiler Optimizations, update ode Generator May, 2 2 22 June, 22 May, 2 2 22 pril, 22 lide # lide #2
PM hart tart ime ctivity 2 Finish dv: hows a graphical flow chart of task dependencies and precedence with times for each task is: oes not show time line clearly lide # lide #4 PM hart xamples???? Not orrect! lide #5 lide #6
PM hart xamples PM hart xamples tart 2 4 OK Finish tart 2 4 Finish lide #7 lide #8 PM hart (Microsoft Project) lide #9 lide #2
PM hart PM ules tart ime ctivity 2 Finish before and and = + + before lide #2 lide #22 PM ummy ctivities PM ummy ctivities U NO! U NO! U Y! U Y! lide #2 lide #24
olution teps for PM Problems PM xample Problem N= N6= tart= N2=4 N4=9 N7= nd=6 Write the earliest start for each node circle. Wait until all input tasks have arrived. Write down all possible paths. o this alphabetically from to Z. Write down the total length of each path including wait times at nodes. Identify the critical path, i.e., the longest path without wait times. Write down, the earliest start for each task activity, this is the time at the node where the task begins. etermine the latest start for each task, L. his is the time at the first node which intersects the critical path minus the path lengths between this ending node and the node at the start of the task minus the time of the node at the start of the task. Write down the latest start for each activity alculate the float for each activity using Float = L -. Paths FJP MP N FJP GJP HIJP HKP HL N=2 N5=6 Lengths +++4w+4+=6 +2++7w+=6 +2+2=5 2+2w++4w+4+=6 2+2+5w+4+=6 2+4++4+=6 2+4+2+5w+=6 2+4+2=8 Float = L - ctivity L Float 4 4 6 6 5 4 8 7 F 4 8 4 G 2 7 5 H 2 2 lide #25 lide #26 P Program valuation and eview echnique ypical ctivity Probability istributions for P harts P charts add time estimates to PM charts t : o Optimistic time estimate. he shortest time the project can be completed kewed-right eta istribution t : m t : p Most likely time required to complete the activity (mode) Pessimistic time estimate. he longest time to complete when everything goes wrong. Murphy s Law kewed-left eta istribution lide #27 lide #28
tart P hart Network iagrams For each path: to, tm, t p 2-5- --5 2-4-7 -- -2-5 Note: No time values for dummy paths Finish t : o t : m t : p t : e t e ompute xpected ask uration Optimistic time Most likely time (Mode) Pessimistic time (Murphy at work!) xpected duration or time [2 t + ( t + t ) / 2] t + 4t + t = = 6 Usually: m o p o m p t e t m lide #29 lide # ompute tandard eviation and Variance Previous P xample tandard deviation: t p to σ = 6 Variance: σ 2 t p to = 6 2 ask xpected ime 4.7 5..7 2. Path ime 7.7 5. 8.8 Project uration = 8.8 lide # lide #2
Previous P xample 4.7 Project uration = 8.8 Path Probability of ime to Project ompletion 5..7 2. What is the probability that the time will be less than any specified time? ask xpected ime 4.7 5..7 2. 5. 6.5 L.66 4.66 5. 6.5 L = + Float FLO.66.66 td. ev..667.8..5.667 Pr( < s ) =? lide # lide #4 ssumption For tandard eviation Of otal ime o ompletion Let the critical path be: 2 2 2 2 with variances σ, σ, σ, σ Α We assume the standard deviation of the total path [ ] is given by: 2 2 2 2 σ = ( σ + σ + σ + σ + ) We assume σ represents a normal distribution / 2 Probability of ime to Project ompletion Pr( < s ) =?. onvert from a normal distribution to a standard normal distribution z s ( e ) = s σ Τ σ Τ = tandard deviation of the total time to completion 2. Use Z table to look up probability lide #5 lide #6
ask Previous P xample 5. td. ev..667.8..5.667.7 4.7 Project uration = 8.8 Path 2. hus for the previous example: 2 2 2 σ = ( σ + σ + σ ) / 2 = (. +.5 +.667 ) =.57 2 2 2 / 2 Probability for ny pecific ime to Project ompletion What is the probability that the time will be greater than any =? Pr( > ) =? onvert random variable to standard variable Z, For z s ( s e ) 8.8 = = =.745.57 σ Τ = Pr( > ) =.228 2% lide #7 lide #8 Posted ail-nd Z able Pr( > ).2 Pr( < ) = -.2 =.77 lide #9 lide #4
eading (Hyman) hapter 7 lide #4