1 / 45

EE384y: Packet Switch Architectures Matchings, implementation and heuristics

EE384y: Packet Switch Architectures Matchings, implementation and heuristics. Nick McKeown Professor of Electrical Engineering and Computer Science, Stanford University nickm@stanford.edu www.stanford.edu/~nickm. Outline. Finding a maximum match. Maximum network flow problems

dava
Télécharger la présentation

EE384y: Packet Switch Architectures Matchings, implementation and heuristics

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. EE384y: Packet Switch Architectures Matchings, implementation and heuristics Nick McKeown Professor of Electrical Engineering and Computer Science, Stanford University nickm@stanford.edu www.stanford.edu/~nickm

  2. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  3. 10 10 10 1 10 1 10 1 Network Flows a c Source s Sink t b d • Let G = [V,E] be a directed graph with capacity cap(v,w) on edge [v,w]. • A flow is an (integer) function, f, that is chosen for each edge so that • We wish to maximize the flow allocation.

  4. 10 10 10 1 10 1 10 1 a c 10, 10 Source s Sink t 10, 10 1 10, 10 10 1 10 b d 1 Flow is of size 10 A maximum network flow exampleBy inspection a c Source s Sink t b d Step 1:

  5. Not obvious Maximum flow: a c 10, 9 Source s Sink t 10, 10 1,1 10, 10 1,1 10, 2 b d 10, 2 1, 1 Flow is of size 10+2 = 12 A maximum network flow example Step 2: a c 10, 10 Source s Sink t 10, 10 1 10, 10 1 10, 1 b d 10, 1 1, 1 Flow is of size 10+1 = 11

  6. Ford-Fulkerson method of augmenting paths • Set f(v,w) = -f(w,v) on all edges. • Define a Residual Graph, R, in which res(v,w) = cap(v,w) – f(v,w) • Find paths from s to t for which there is positive residue. • Increase the flow along the paths to augment them by the minimum residue along the path. • Keep augmenting paths until there are no more to augment.

  7. Example of Residual Graph a c 10, 10 10, 10 1 10, 10 s t 10 1 10 b d 1 Flow is of size 10 Residual Graph, R res(v,w) = cap(v,w) – f(v,w) a c 10 10 10 1 s t 10 1 10 b d 1 Augmenting path

  8. Example of Residual Graph Step 2: a c 10, 10 s t 10, 10 1 10, 10 1 10, 1 b d 10, 1 1, 1 Flow is of size 10+1 = 11 Residual Graph a c 10 s t 10 10 1 1 1 1 b d 9 1 9

  9. Example of Residual Graph Step 3: a c 10, 9 s t 10, 10 1, 1 10, 10 1, 1 10, 2 b d 10, 2 1, 1 Flow is of size 10+2 = 12 1 Residual Graph a c 9 s t 10 10 1 2 1 2 b d 8 1 8

  10. Complexity of network flow problems • In general, it is possible to find a solution by considering at most |V|.|E| paths, by picking shortest augmenting path first. • There are many variations, such as picking most augmenting path first.

  11. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  12. A 1 2 B 3 C 4 D 5 E 6 F Finding a maximum size match • How do we find the maximum size (weight) match?

  13. Network flows and bipartite matching A 1 2 B Finding a maximum size bipartite matching is equivalent to solving a network flow problem with capacities and flows of size 1. Sink t Source s 3 C 4 D 5 E 6 F

  14. Network flows and bipartite matchingFord-Fulkerson method Residual Graph for first three paths: A 1 2 B t s 3 C 4 D 5 E 6 F

  15. Network flows and bipartite matching Residual Graph for next two paths: A 1 2 B t s 3 C 4 D 5 E 6 F

  16. Network flows and bipartite matching Residual Graph for augmenting path: A 1 2 B t s 3 C 4 D 5 E 6 F

  17. Network flows and bipartite matching Residual Graph for last augmenting path: A 1 2 B t s 3 C 4 D 5 E 6 F Note that the path augments the match: no input and output is removed from the match during the augmenting step.

  18. Network flows and bipartite matching Maximum flow graph: A 1 2 B t s 3 C 4 D 5 E 6 F

  19. Network flows and bipartite matching Maximum Size Matching: A 1 2 B 3 C 4 D 5 E 6 F

  20. Complexity of Maximum Matchings • Maximum Size Matchings: • Algorithm by Dinic O(N5/2) • Maximum Weight Matchings • Algorithm by Kuhn O(N3) • In general: • Hard to implement in hardware • Slooooow.

  21. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  22. Maximal Matching • A maximal matching is one in which each edge is added one at a time, and is not later removed from the matching. • i.e. no augmenting paths allowed (they remove edges added earlier). • No input and output are left unnecessarily idle.

  23. A 1 A 1 A 1 2 B 2 B 2 B 3 C 3 C 3 C 4 D 4 4 D D 5 E 5 5 E E 6 F 6 6 F F Example of Maximal Size Matching Maximal Size Matching Maximum Size Matching

  24. Maximal Matchings • In general, maximal matching is simpler to implement, and has a faster running time. • A maximal size matching is at least half the size of a maximum size matching. • A maximal weight matching is defined in the obvious way. • A maximal weight matching is at least half the weight of a maximum weight matching.

  25. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  26. Wave Front Arbiter(Tamir) Requests Match 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4

  27. Wave Front Arbiter Requests Match

  28. 2,4 1,4 4,4 3,4 4,3 4,2 4,1 3,3 3,2 3,1 2,2 2,1 1,3 1,2 1,1 2,3 Wave Front ArbiterImplementation Simple combinational logic blocks

  29. Wave Front ArbiterWrapped WFA (WWFA) N steps instead of 2N-1 Match Requests

  30. Wavefront ArbitersProperties • Feed-forward (i.e. non-iterative) design lends itself to pipelining. • Always finds maximal match. • Usually requires mechanism to prevent Q11 from getting preferential service. • In principle, can be distributed over multiple chips.

  31. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  32. Iteration: 1 1 1 1 2 2 2 2 #1 1 1 3 3 3 3 2 2 4 4 4 4 f2: Grant f3: Accept/Match 3 3 1 1 1 1 1 1 4 4 2 2 2 2 2 2 #2 3 3 3 3 3 3 4 4 4 4 4 4 Parallel Iterative Matching uar selection uar selection f1: Requests

  33. PIM Properties • Guaranteed to find a maximal match in at most N iterations. • In each phase, each input and output arbiter can make decisions independently. • In general, will converge to a maximal match in < N iterations. • How many iterations should we run?

  34. Parallel Iterative MatchingConvergence Time Number of iterations to converge:

  35. Parallel Iterative Matching

  36. Parallel Iterative Matching PIM with a single iteration

  37. Parallel Iterative Matching PIM with 4 iterations

  38. Outline • Finding a maximum match. • Maximum network flow problems • Definitions and example • Augmenting paths • Maximum size/weight matchings as examples of maximum network flows • Maximum size matching • Complexity of maximum size matchings and maximum weight matchings • What algorithms are used in practice? • Maximal Matches • Wavefront Arbiter (WFA) • Parallel Iterative Matching (PIM) • iSLIP

  39. 1 4 2 1 1 1 1 3 2 2 2 2 #1 1 1 3 3 3 3 2 2 4 4 4 4 F2: Grant F3: Accept/Match 3 3 1 1 1 1 1 1 4 4 2 2 2 2 2 2 #2 3 3 3 3 3 3 4 4 4 4 4 4 iSLIP F1: Requests

  40. iSLIP Operation • Grant phase: Each output selects the requesting input at the pointer, or the next input in round-robin order. It only updates its pointer if the grant is accepted. • Accept phase: Each input selects the granting output at the pointer, or the next output in round-robin order. • Consequence: Under high load, grant pointers tend to move to unique values.

  41. iSLIPProperties • Random under low load • TDM under high load • Lowest priority to MRU • 1 iteration: fair to outputs • Converges in at most N iterations. (On average, simulations suggest < log2N) • Implementation: N priority encoders • 100% throughput for uniform i.i.d. traffic. • But…some pathological patterns can lead to low throughput.

  42. iSLIP

  43. iSLIP

  44. iSLIPImplementation Programmable Priority Encoder 1 State 1 log2N Decision N Grant Accept 2 2 N Grant Accept log2N N N Grant Accept log2N N

  45. Maximal Matches • Maximal matching algorithms are widely used in industry (PIM, iSLIP, WFA and others). • PIM and iSLIP are rarely run to completion (i.e. they are sub-maximal). • We will see shortly that a maximal match with a speedup of 2 is stable for non-uniform traffic.

More Related