Download presentation
Presentation is loading. Please wait.
Published byDomenic Nichols Modified over 9 years ago
1
Minimizing Flow Time on Multiple Machines Nikhil Bansal IBM Research, T.J. Watson
2
Scheduling Collection of m machines, n jobs Arrival time or release time (r j ) Service requirement or size (p j ) t=0 r2r2 r3r3 r1r1 C2C2 C1C1 Job preempted C3C3 m=1
3
Scheduling Flow Time = Time job spends = Completion time – release time = Waiting + Processing t=0 r2r2 r3r3 r1r1 C2C2 C1C1 C3C3 Flow time of job 2 m=1
4
Scheduling t=0 r2r2 r3r3 r1r1 C2C2 C1C1 C3C3 Flow time of job 3 Flow Time = Time job spends = Completion time – release time = Waiting + Processing minimize total flow time m=1
5
Total Flow Time (Another View) Imagine each job costs $1 per unit time. Cost of a job = Its flow time Total cost = Total flow time Total cost = t cost at time t = t # jobs at time t
6
Total Flow Time (Single Machine) Total cost = t # jobs at time t Processor has a “to do” list of jobs Goal: Minimize number of jobs on list Work on the job it can finish earliest. Shortest remaining processing time (SRPT): Optimal algorithm
7
Flow Time on multiple Machines (m ¸ 2) NP-Hard: Breakthrough: O(log n) competitive [Leonardi, Raz 97] Works for arbitrary # of machines (m) Any online algorithm: (log n) competitive Improvements: No migrations [Awerbuch et al 99] Immediate dispatch [Avrahami and Azar 03]
8
Flow Time on Multiple Machines What about approximation algorithms? O(log n) best known, even for m=2 Lower bounds: NP-Hard, APX-Hard ?
9
Flow Time on Multiple Machines Main Result: A (1+) approximation scheme Running Time = n O(m log n) Or, n O(log n) for m=O(1) Suggests: PTAS likely for O(1) machines
10
Basic Idea Rounding: Simplify the input without losing quality too much Search: Dynamic Programming over some reasonable space of schedules
11
Related Problem Minimizing total completion time: ( i c i or equivalently i (r i + f i ) ) Same as flow time wrt optimality But easier for approximation PTASes known with runtime poly(n,m) [Afrati et al 99] Techniques not applicable to flow time
12
Rounding for Flow Time Flow Time is quite sensitive Suppose round size to powers of (1+) Cannot distinguish between Job of size 1 arrives at t=1,2,…,n Job of size 1+ arrives at t=1,2,…,n Very Different: (n) vs (n 2 ) !!!
13
Rounding for Flow Time Can show: Let B be largest size, Rounding r i, p i to multiples of B/n 2 is fine Proof: Each job affected by · B/n Opt ¸ B Implies: Sizes 2 [1,n 2 /], Events at [1,n 3 /] Still bad for exhaustive search over all schedules.
14
Restricting possible schedules Jobs assigned to a machine, worked in SRPT order. Given a machine, which jobs assigned to it? (2 n possibilities) Approx state under SRPT in O(log 2 n) bits of info. Store for each machine. Dynamic program: For (state,t) whats the best flow time achievable.
15
State Properties 1) Enough information: State at t+1 computable from that at time t. 2) Gives number of jobs to within 1+ factor
16
Property of SRPT At any time, among jobs with size 2 [a,b], at most one has remaining processing < a.
17
Property of SRPT At any time, among jobs with size 2 [a,b], at most one has remaining processing < a. Proof: b a Not executed until blue finishes
18
Property of SRPT At any time, among jobs with size 2 [a,b], at most one has remaining processing < a. Proof: b a Both cannot be < a at some time
19
Property of SRPT At any time, among jobs with size 2 [a,b], at most one has remaining processing < a. Suppose a= (1+) i, b=(1+) i+1 Given, total remaining size (x) of jobs s.t. p i 2 [a,b] x/b · Estimate # of jobs · x/a + 1
20
Configuration on a machine Consider O(log n/) size-classes [(1+) i,(1+) i+1 ] For each class, Total remaining processing times 1/ largest remaining processing times x/(1+) i+1 · # of jobs · x/(1+ i ) + 1 Class 1: (Total 1, x 1,x 2,…,x 1/ ) … Class k: (Total k, y 1,y 2,…,y 1/ ) k=O(log n) In all O(log 2 n) bits
21
Updating a configuration At most O(m log 2 n) bits of information Gives number of jobs to within 1+ How to update, as time passes? Class 1: (Total 1, x 1,x 2,…,x 1/ ) … Class j : (Total j, y 1,y 2,…,y 1/ ) On arrival, guess the machine & update state m branches
22
Updating a configuration At most O(m log 2 n) bits of information Gives number of jobs to within 1+ How to update, as time passes? Class 1: (Total 1, x 1,x 2,…,x 1/ ) … Class j : (Total j, y 1,y 2,…,y 1/ ) Working step: For each machine, guess class with smallest remaining time job [(log n) m choices]
23
Fitting it all together At any time, O(m log 2 n/ 2 ) total bits of info. Know how to update. Dynamic program over all possible states.
24
Weighted Flow Time ( i w i f i ) NP-Hard for m=1, No o(n) approximation known, even for m=1 m=1: (1+) approx, time n O(log B log W) [Chekuri, Khanna 02] B: max/min size W: max/min wt This paper: Extend to m=O(1), time n O(m log Bn log Wn) Hardness: Exponential dependence on m likely (1+ ) approx with running time 2 O(polylog(n,m,W,B)) ) NP µ DTIME(n polylog(n) )
25
Open Problems 1) PTAS or O(1) approx for minimizing flow time on O(1) machines? [Our QPTAS => PTAS likely] 2) For arbitrary number of machines. PTAS or APX-Hard?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.