ISPD 2008 Global Routing Contest Cliff Sze, Gi-Joon Nam, Mehmet - - PDF document

▶
ispd 2008 global routing contest
SMART_READER_LITE
LIVE PREVIEW

ISPD 2008 Global Routing Contest Cliff Sze, Gi-Joon Nam, Mehmet - - PDF document

Jan 31, 2024 134 likes •410 views

ISPD 2008 Global Routing Contest Cliff Sze, Gi-Joon Nam, Mehmet Yildiz IBM Corp 1 Overview Open contest primarily for academic community Totally 11 team registered initially All academic teams 4 teams from US, 7 teams

slide-1
SLIDE 1

1

ISPD 2008 Global Routing Contest

Cliff Sze, Gi-Joon Nam, Mehmet Yildiz IBM Corp

2

Overview

  • Open contest primarily for academic community
  • Totally 11 team registered initially
  • All academic teams
  • 4 teams from US, 7 teams from overseas
  • 1 team Hannover, Germery
  • 4 from Taiwan
  • 2 from Hong Kong
  • 4 new participations
  • 10 final entries
  • Total 16 benchmarks
  • 8 from 2007 global routing contest “3D” benchmarks
  • 8 new global routing benchmarks are released
  • All derived from ISPD 2005/ 2006 placement benchmark

solutions

  • Quality metrics
  • Minimizing overflows
  • CPU-weighted total wirelength
slide-2
SLIDE 2

3

New for 2008

  • CPU time is restricted to 24 hours
  • Any run more than 24 hours = > fail
  • CPU-weighted total wirelength
  • Parallel Algorithm is allowed (at most 4 CPU is allowed)
  • Review our metrics from last year
  • G-cell size is good
  • After reviewing the data of several technology generations, we set via cost = 1 g-

cell ( the resistance ratio between via is slightly lower than one unit of wire in one g-cell). It was set to 3 last year.

  • Thanks to our “Consultants”
  • Emails Discussions on the “best” metric
  • Patrick Groeneveld - Magma
  • Prashant Saxena – Synopsys
  • Jeffery Salowe - Cadence
  • Philip Chong – Cadence
  • Mustafa Ozdal – Intel
  • Gustavo Tellez - IBM
  • Stephen Quay - IBM
  • Good Metrics
  • Total overflow
  • Maximum overflow
  • # of nets with overflow
  • Average 20% worse congestion nets

4

Quality Metrics

  • Final quality metric
  • Minimum Σ Rank(circuit)

wins the game

  • Rank per circuit is

determined by

  • Minimum total
  • verflows
  • Max overflow as the

1st tie breaker

  • Routed wire length as

the 2nd tie breaker

  • Routed wire length calculation

considers via cost

  • One via connecting two

consecutive metal layer = WL of one g-cell

  • CPU-weighted wirelength

Example from Mustafa Ozdal, Intel. Corp.

slide-3
SLIDE 3

5

CPU-Weighted Wire Length Calculation

  • 2nd tie-breaker
  • routed_wire_length * (1 +

CPU_time_factor)

  • CPU_time_factor =

0.04 log2( router_cpu_time / median_cpu_time)

  • CPU_time_factor will range

from -0.1 to 0.1

  • max 10%

routed_wire_length advantage or disadvantage

  • if a router is 2x

faster/ slower, the router gets about 4% routed wire length advantage/ disadvantage

  • Similar to the one used in the

ISPD 2006 Placement Contest

6

How Benchmarks were Generated

  • For each ISPD 2005/ 2006 benchmark
  • adaptec1, adaptec2, adaptec3, adaptec4, adaptec5,

newblue1, newblue2, newblue3 (2007)

  • bigblue1, bigblue2, bigblue3, bigblue4, newblue4,

newblue5, newblue6, newblue7 (2008) 1. Pick Placement tools

  • Capo, mPL6, Dragon, APlace3, mFAR, NTUPlace3.0,

FastPlace3.0, Kraftwerk 2. Pick density target

  • From 50% to 90%

3. Generate placement solution 4. Impose a tile structure

  • Basic routing resources are determined

5. Adjust routing resources

slide-4
SLIDE 4

7

2000 4000 6000 8000 10000 12000 2000 4000 6000 8000 10000 12000

newblue1 #Cells= 330474, #Nets= 331663

* Placement layout figure was generated by Capo Placer utility package. 8

slide-5
SLIDE 5

9

Albrecht, C. TCAD 2001 paper on multicommodity flow based global routing algorithm.

10

Routing Resource (Edge Capacity) Adjustment

  • Essentially determines the level of difficulty of benchmark
  • Tile size
  • 30–50 wire tracks
  • Limited usage in M1/ M2 layer
  • 20% of available wire tracks
  • Guard band
  • 90-100% of tile size
  • Blockage Porosity

adaptec3.dragon70.3d.30.50.90.gr

ISPD Placement Benchmark Name

slide-6
SLIDE 6

11

Edge Capacity Adjustment

  • Tile size
  • 30–50 wire tracks
  • Limited usage in M1/ M2 layer
  • 20% of available wire tracks
  • Guard band
  • 90-100% of tile size
  • Blockage Porosity

adaptec3.dragon70.3d.30.50.90.gr

Placement tool used and its density target Tile size Block Porosity Guard band

12

More on Block Porosity

  • Affects any tiles that sit on top
  • f blockages
  • Only affects M3/ M4 metal

layers

2000 4000 6000 8000 10000 12000 2000 4000 6000 8000 10000 12000

newblue1 #Cells= 330474, #Nets= 331663

slide-7
SLIDE 7

13

Overview of all 16 benchmarks

16284051 1647410 8 (4/4) 20 82 80 488 x 490 80

KraftWerk

newblue7*

n7

8823094 835267 6 (3/3) 10 100 60 463 x 464 80

mFAR

newblue6*

n6

12357104 891920 6 (3/3) 10 100 40 637 x 640 50

NTUplace

newblue5*

n5

7357235 531292 6 (3/3) 10 95 40 455 x 458 50

mPL6

newblue4*

n4

6998467 442005 6 (3/3) 50 90 40 973 x 1256 80

KraftWerk

newblue3

n3

4191219 373790 6 (3/3) 20 100 50 557 x 463 90

FastPlace

newblue2

n2

2079947 270713 6 (3/3) 50 90 30 399 x 399 80

NTUplace

newblue1

n1

10489255 1133535 8 (4/4) 20 80 80 403 x 405 70

FastPlace

bigblue4*

b4

7170444 665629 8 (4/4) 10 90 50 555 x 557 70

APlace

bigblue3*

b3

4049521 428968 6 (3/3) 60 60 40 468 x 471 60

mPL6

bigblue2*

b2

2986719 196885 6 (3/3) 10 100 50 227 x 227 60

Capo

bigblue1*

b1

8896706 548073 6 (3/3) 20 100 50 465 x 468 50

mFAR

adaptec5

a5

8175006 401060 6 (3/3) 50 90 30 774 x 779 60

APlace

adaptec4

a4

8619596 368494 6 (3/3) 50 90 30 774 x 779 70

Dragon

adaptec3

a3

2882254 207972 6 (3/3) 20 100 35 424 x 424 60

mPL6

adaptec2

a2

3000320 176715 6 (3/3) 50 90 35 324 x 324 70

Capo

adaptec1

a1

HPWL #Nets Layers Porosity

  • band

Size Dimension Target Solution Circuit Tile #Metal Macro Guard Tile Tile Density Placement

14

  • ALL teams!!!!
  • I truly think all teams should be the winner because…
  • Most results are better than last-year’s best-results
  • n1 is routable by a few routers
  • n3 is provably unroutable

ISPD 2008 Global Routing Contest Winner

4.12% 60.87%

  • 2.40%

109.17 414 33627 MaizeRouter 113.86 1058 32840 n3 0.89% na na 76.86 MaizeRouter 77.55 n2 3.35% 0.00% 89.00% 48.98 2 44 BoxRouter 50.68 2 400 n1 5.74% na na 160.38 BoxRouter 170.14 a5 1.71% na na 122.59 MaizeRouter 124.72 a4 2.09% na na 133.43 MaizeRouter 136.27 a3 5.20% na na 53.08 MaizeRouter 55.99 a2 5.05% na na 56.52 MaizeRouter 59.52 a1 (e5) Max. Total (e5) Max. Total Router (e5) Max. Total Total WL Overflow Total WL Overflow Total WL Overflow Improvement from 2007 2008 median 2007 BEST results

slide-8
SLIDE 8

15

Let me take a few minutes to introduce all teams

  • I asked the teams to send me
  • Names, affiliation
  • Photos
  • And …

. of course a brief description of their router

16

Team 1 – FastRoute 3.0

  • Iowa State University VLSI CAD LAB
  • Yanheng Zhang, Yue Xu
  • Advisor: Dr. Chris Chu
  • Description
  • 1. Initial Congestion Map

Generation.

  • 2. Use FLUTE to generate initial

RSMT.

  • 3. Generate Congestion Driven

RSMT.

  • 4. Via guided Pattern Routing.
  • 5. Maze Routing until best
  • verflow
  • 6. Layer Assignment
  • Router uses at most 4 CPU
slide-9
SLIDE 9

17

Team 2 – FGR

  • University of Michigan
  • Jarrod Roy
  • Advisor: Dr. Igor Markov
  • Description
  • 1) Decompose nets by Minimum Spanning

Tree

  • 2) Initial routing and rip-up and reroute

using an A* -driven maze router

  • 3) Rip-up and reroute using an A* -driven

maze router and discrete lagrange multipliers

  • 4) Net topology reconfiguration during rip-up

and reroute via epsilon-sharing

  • 5) Fast layer assignment
  • 6) full 3-D maze routing greedy cleanup to

recover wirelength

  • Router uses 1 CPU

18

Team 3 – NTUgr

  • National Taiwan University
  • Huang-Yu Chen, Chin-Hsiung Hsu
  • Advisor: Dr. Yao-Wen Chang
  • Description
  • 1. Prerouting with high-pin density analysis
  • 2. Initial iterative monotonic routing
  • 3. Enhanced iterative negotiation-based rip-up/ rerouting

(a) ultra-fast rerouting selection (b) parallel routing speed up

  • 4. Parallel layer assignment
  • Router uses at most 4 CPU
slide-10
SLIDE 10

19

Team 4 – NCTU

  • National Chiao Tung University
  • Wen-Hao Liu, Ke-Ren Dai
  • Advisor: Yih-Lang Li
  • Description
  • 1. Use minimum spanning tree

to generate 2-pin connection for each net

  • 2. Use monotonic routing to get

initial routing

  • 3. Do evolution-based rip-up

and reroute with historical cost to get 2D global routing result

  • 4. Do layer assignment to

complete 3D global routing

  • Router uses 1 CPU

20

Team 5 – BoxRouter

  • University of Texas - Austin
  • Minsik Cho, Katrina Lu, Kun Yuan
  • Advisor: Dr. David Pan
  • Description
  • 1. Prerouting with flat routing.
  • 2. Use FLUTE2.5 for net

decomposition.

  • 3. Initial routing and rerouting

with maze routing. Use history-based method.

  • 4. Layer assignment, starting

with nets with short WL/ small number of pins.

  • Router uses 1 CPU
slide-11
SLIDE 11

21

Team 6 – NTHU-Route

  • National Tsing Hua University, Taiwan
  • Yen-Jung Chang, Yu-Ting Lee, Tsung-Hsien Lee
  • Advisor: Dr. Ting-Chi Wang
  • Description
  • 1. Project the design onto a plane.
  • 2. Decompose multi-pin nets by FLUTE.
  • 3. Route multi-pin nets with probabilistic L-shaped

pattern routing to get initial congestion map.

  • 4. Change tree topologies of multi-pin nets with edge

shifting.

  • 5. Route multi-pin with L-shaped pattern routing.
  • 6. Rip-up and reroute by congested region
  • identification. The routing-techniques are monotonic

routing and adaptive multi-source multi-sink maze routing.

  • 7. Repeat step 4. until one of the situation is

satisfied:

  • verflow under a threshold.
  • iteration count reach the maximum iteration

count limitation.

  • reduction of overflow is suspend too long.
  • some others...
  • 8. Refine the overflowed two-pin net by the same

routing technique in step 4. but applied with different cost function.

  • 9. Layer assignment.
  • Router uses 1 CPU

22

Team 7 – IMS GlObal Router (IGOR)

  • Institute of Microelectronic Systems(IMS),

Hannover, Germany

  • Artur Quiring, Philipp Panitz, Ole Ohlendorf
  • Description
  • 1. Use Flute2.5 to get a RSMT for each net
  • 2. Use the Steiner tree information and

route each net with maze routing starting with smallest net.

  • 3. Rip up and reroute with maze routing,

using different cost function then in step 2

  • starting with longest route with overflow
  • try to reroute the route with no overflow

(edges with overflow are not considered)

  • if not possible, try to reroute the route

and accept new route if overflow is reduced (edges with overflow are considered)

  • Router uses 1 CPU
slide-12
SLIDE 12

23

Team 9 – HSR (Heuristically Statistical Router)

  • Department of Electronics Engineering,

National Chiao Tung University in Taiwan

  • Sean Liu, Jerry Lee, Po-Cheng Pan, Ching-

Yu Chin, Yi-Hung Chen

  • Advisor: Dr. Hung-Ming Chen
  • Description
  • 1. Sort each net according its density

(standard deviation from pins).

  • 2. Invoke FLUTE2.5 to get a RSMT for

each net

  • 3. for each net, perform quick 3D

pattern route.

  • 4. Analyze the congestion and put
  • verflow edges into an RRHeap.
  • 5. For each edges in RRHeap, reRoute

it with overflow level 0.

  • 6. For each edges in RRHeap, reRoute

it with overflow level -1.

  • 7. For each edges in RRHeap, reRoute

it with overflow level INT_MIN.

  • Router uses 1 CPU
  • First time participation

24

Team 10 – Simple Router

  • Dept of Electronic & Information Engineering,

Polytechnic University of Hong Kong

  • Jingwei Lu
  • Description
  • 1. Use FLUTE2.5 to get a RSMT for each net
  • 2. Compress 3D to 2D and construct a

congestion map in terms of pattern routes.

  • 3. do pattern routing for each net and maze

routing when some wire passes congested area.

  • 4. Remove extra Steiner points and

segments when each net gets routed.

  • 5. Rip-up Reroute with three steps of

different sequence: pattern route sequence, edge sorted by overflow and also each net sorted by overflow.

  • Router uses 1 CPU
  • First time participation
slide-13
SLIDE 13

25

Team 11 – Amaze Global Router (AMGR)

  • Dept of Computer Science and

Engineering, Chinese University of Hong Kong

  • Xiao Linfu, Li Liang, Qian Zaichen
  • Advisor: Dr. Evan Young
  • Description
  • 1. Use our fast multi-pin maze

routing engine repetitively to generate a very good initial global routing solution.

  • 2. Rip-up and re-route with our

engine.

  • 3. Do wire length minimization

and bending reduction.

  • 4. Do layer assignment, starting

with nets with larger number of pins and shorter lengths

  • Router uses 1 CPU
  • First time participation

26

Summary: Binary of routers

46864 1 CUHKrouter-new AMGR 11 2879334 POWV9.dat 6562490 POST9.dat 434384 1 simple_router simple router 10 2879334 POWV9.dat 6562490 POST9.dat 1204578 1 GR HSR 9 2879334 POWV9.dat 6562490 POST9.dat 138316 1 main IGOR 7 2879334 POWV9.dat 6562490 POST9.dat 1489864 1 nthuroute NTHU-Route 6 2879334 POWV9.dat 6562490 POST9.dat 1228160 1 br_ispd08.x.s BoxRouter 5 74479 LA 313094 1 dory NCTU 4 1088443 4 NTUGR NTUGR 3 1382128 1 FGR.exe FGR 2 2879334 POWV9.dat 6562490 POST9.dat 418304 4 FastRoute FastRoute 3.0 1 size

  • ther file 2

size

  • ther file 1

size CPU Binary name Team Name ID

slide-14
SLIDE 14

27

New Benchmarks

  • We adjust our new

benchmarks such that the best open-source routers from 2007 produce small amount of overflow

  • 5 out of 8 is routable by

“median router”

  • We look forward …
  • The 3 currently

unroutable benchmarks will become routable within months

X X 356.08 3 1023 n7 O O 184.67 n6 O O 238.00 n5 X X 139.20 3 222 n4 X X 109.17 414 33627 n3 O O 76.86 n2 O X 48.98 2 44 n1 X X 237.22 4 389 b4 O O 132.02 b3 O O 96.46 b2 O O 59.89 b1 O O 160.38 a5 O O 122.59 a4 O O 133.43 a3 O O 53.08 a2 O O 56.52 a1 routable? routable? (e5) Max. Total router router Total WL Overflow best median 2008 median 28

ISPD 2008 Global Routing Contest Winner

9.31 9 9 9 10 8 10 9 9 9 9 10 10 10 9 9 10 J 9.13 9 9 9 9 9 8 10 9 9 10 9 9 9 9 10 9 I 8.13 8 8 8 8 9 9 8 8 8 8 8 8 8 8 8 8 H 6.50 7 7 7 5 7 7 7 7 7 4 5 7 7 7 7 6 G 5.50 6 6 6 3 6 6 4 4 6 6 7 5 6 5 5 7 F 5.13 5 5 5 7 5 5 3 5 5 3 6 6 5 6 6 5 E 3.63 2 4 2 4 4 4 5 6 2 1 4 4 4 4 4 4 D 3.06 4 2 3 6 3 1 6 3 3 7 2 2 2 1 2 2 C 2.81 3 3 4 1 1 3 2 2 4 5 3 3 3 2 3 3 B 1.38 1 1 1 2 2 2 1 1 1 2 1 1 1 3 1 1 A avgRank n7 n6 n5 n4 n3 n2 n1 b4 b3 b2 b1 a5 a4 a3 a2 a1 x x x v x 2008 2008

  • First time participating teams perform not as good
  • nly have 2 weeks to try 8 new benchmarks
slide-15
SLIDE 15

29

ISPD 2008 Global Routing Contest Winner

  • With CPU-

weighted TWL

  • With TWL

instead of CPU-weighted TWL

9.31 9 9 9 10 8 10 9 9 9 9 10 10 10 9 9 10 J 9.13 9 9 9 9 9 8 10 9 9 10 9 9 9 9 10 9 I 8.13 8 8 8 8 9 9 8 8 8 8 8 8 8 8 8 8 H 6.69 7 7 7 5 7 7 7 7 7 4 7 7 7 7 7 7 G 4.75 6 4 4 3 6 2 4 4 5 6 6 5 5 6 4 6 F 4.75 4 6 5 6 3 5 6 3 3 7 5 6 4 4 5 4 C 4.19 3 5 6 1 1 6 2 2 6 5 4 4 6 5 6 5 B 3.38 5 3 3 7 5 3 3 5 4 3 3 2 1 2 2 3 E 3.00 2 2 2 4 4 4 5 6 2 1 2 3 3 3 3 2 D 1.25 1 1 1 2 2 1 1 1 1 2 1 1 2 1 1 1 A avgRank n7 n6 n5 n4 n3 n2 n1 b4 b3 b2 b1 a5 a4 a3 a2 a1 9.31 9 9 9 10 8 10 9 9 9 9 10 10 10 9 9 10 J 9.13 9 9 9 9 9 8 10 9 9 10 9 9 9 9 10 9 I 8.13 8 8 8 8 9 9 8 8 8 8 8 8 8 8 8 8 H 6.50 7 7 7 5 7 7 7 7 7 4 5 7 7 7 7 6 G 5.50 6 6 6 3 6 6 4 4 6 6 7 5 6 5 5 7 F 5.13 5 5 5 7 5 5 3 5 5 3 6 6 5 6 6 5 E 3.63 2 4 2 4 4 4 5 6 2 1 4 4 4 4 4 4 D 3.06 4 2 3 6 3 1 6 3 3 7 2 2 2 1 2 2 C 2.81 3 3 4 1 1 3 2 2 4 5 3 3 3 2 3 3 B 1.38 1 1 1 2 2 2 1 1 1 2 1 1 1 3 1 1 A avgRank n7 n6 n5 n4 n3 n2 n1 b4 b3 b2 b1 a5 a4 a3 a2 a1 x x x v x

30

ISPD 2008 Global Routing Contest Winner

9.31 9 9 9 10 8 10 9 9 9 9 10 10 10 9 9 10 J 9.13 9 9 9 9 9 8 10 9 9 10 9 9 9 9 10 9 I 8.13 8 8 8 8 9 9 8 8 8 8 8 8 8 8 8 8 H 6.50 7 7 7 5 7 7 7 7 7 4 5 7 7 7 7 6 G 5.50 6 6 6 3 6 6 4 4 6 6 7 5 6 5 5 7 F 5.13 5 5 5 7 5 5 3 5 5 3 6 6 5 6 6 5 E 3.63 2 4 2 4 4 4 5 6 2 1 4 4 4 4 4 4 D 3.06 4 2 3 6 3 1 6 3 3 7 2 2 2 1 2 2 C 2.81 3 3 4 1 1 3 2 2 4 5 3 3 3 2 3 3 B 1.38 1 1 1 2 2 2 1 1 1 2 1 1 1 3 1 1 A avgRank n7 n6 n5 n4 n3 n2 n1 b4 b3 b2 b1 a5 a4 a3 a2 a1 x x x v x 2008 2008

AMGR BoxRouter FGR FastRoute 3.0 NTUgr NTHU-route

slide-16
SLIDE 16

31

How does 2008-best compare to 2007-best

6.47% 80.72% 4.22% NTHU-route 106.49 204 31454 MaizeRouter 113.86 1058 32840 n3 2.37% na na NTHU-route 75.71 MaizeRouter 77.55 n2 8.19% WOW WOW NTHU-route 46.53 BoxRouter 50.68 2 400 n1 8.57% na na NTHU-route 155.55 BoxRouter 170.14 a5 2.40% na na FGR 121.73 MaizeRouter 124.72 a4 10.72% na na NTHU-route 121.66 MaizeRouter 136.27 a3 6.57% na na NTHU-route 52.31 MaizeRouter 55.99 a2 10.12% na na NTHU-route 53.50 MaizeRouter 59.52 a1 (e5) Max. Total Router (e5) Max. Total Router (e5) Max. Total Total WL Overflow Total WL Overflow Total WL Overflow Improvement from 2007 2008 BEST results 2007 BEST results 32

Summary

  • Academic routers show great improvements
  • A good set of global routing benchmarks

Overflow minimization Routed wire length minimization, CPU-weighted

  • Future directions?
  • What is a good/ fast way to evaluate global routing?

Information inside a g-cell is ignored

  • Look forward to next global/ detailed routing contest
  • Open to any suggestions/ feedbacks
  • http: / / www.ispd.cc/ rcontest
slide-17
SLIDE 17

33

ISPD 2008 Global Routing Contest Detailed Results

34

Overall Rankings

9.31 9 9 9 10 8 10 9 9 9 9 10 10 10 9 9 10 J HSR 9.13 9 9 9 9 9 8 10 9 9 10 9 9 9 9 10 9 I Simple Router 8.13 8 8 8 8 9 9 8 8 8 8 8 8 8 8 8 8 H IGOR 6.50 7 7 7 5 7 7 7 7 7 4 5 7 7 7 7 6 G AMGR 5.50 6 6 6 3 6 6 4 4 6 6 7 5 6 5 5 7 F NCTU 5.13 5 5 5 7 5 5 3 5 5 3 6 6 5 6 6 5 E FGR 3.63 2 4 2 4 4 4 5 6 2 1 4 4 4 4 4 4 D BoxRouter 3.06 4 2 3 6 3 1 6 3 3 7 2 2 2 1 2 2 C FastRoute 3.0 2.81 3 3 4 1 1 3 2 2 4 5 3 3 3 2 3 3 B NTUgr 1.38 1 1 1 2 2 2 1 1 1 2 1 1 1 3 1 1 A NTHU-Route avgRank n7 n6 n5 n4 n3 n2 n1 b4 b3 b2 b1 a5 a4 a3 a2 a1 x x x v x 2008 2008

slide-18
SLIDE 18

35

a1

  • - : cannot finish in 24 hours, or router crashes

10 0.100 8786880 11203 7988073 216 671842 HSR 9 0.082 8108653 6863 7495873 12 2974 Simple Router 8 0.100 12440804 34112 11309822 IGOR 6

  • 0.018

5706498 1209 5813658 AMGR 7 0.018 5796210 2274 5693690 NCTU 5 0.013 5477954 2102 5405141 FGR 4

  • 0.018

5285899 1227 5380627 BoxRouter 2

  • 0.100

4998691 105 5554101 FastRoute 3.0 3

  • 0.100

5048564 293 5609515 NTUgr 1

  • 0.072

4962279 475 5349595 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 36

a2

  • - : cannot finish in 24 hours, or router crashes

9 0.100 8321233 22208 7564757 100 274474 HSR 10

  • Simple Router

8 0.100 12568304 52005 11425731 6 824 IGOR 7 0.053 5853963 1585 5558821 AMGR 5 0.000 5274884 632 5274884 NCTU 6 0.017 5347743 855 5255902 FGR 4

  • 0.079

4857235 162 5271806 BoxRouter 2

  • 0.100

4777025 25 5307805 FastRoute 3.0 3

  • 0.100

4803787 73 5337541 NTUgr 1

  • 0.100

4707881 104 5230979 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-19
SLIDE 19

37

a3

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.100 29197430 41933 26543118 IGOR 7 0.044 14678128 3779 14061475 AMGR 5 0.004 13453456 1900 13397630 NCTU 6 0.037 13654190 3327 13173349 FGR 4

  • 0.004

13119867 1635 13179058 BoxRouter 1

  • 0.100

11997446 101 13330495 FastRoute 3.0 2

  • 0.100

12020225 309 13355806 NTUgr 3

  • 0.075

12126960 481 13111262 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 38

a4

  • - : cannot finish in 24 hours, or router crashes

10 0.100 18342586 75094 16675078 54 145016 HSR 9 0.100 20162227 68823 18329297 4 22 Simple Router 8 0.100 22942918 7132 20857198 IGOR 7 0.062 13587555 2756 12790383 AMGR 6

  • 0.004

12210719 876 12257582 NCTU 5 0.004 12210400 996 12166776 FGR 4

  • 0.049

11618867 403 12211779 BoxRouter 2

  • 0.100

11000618 33 12222909 FastRoute 3.0 3

  • 0.100

11033537 104 12259485 NTUgr 1

  • 0.100

10955423 135 12172692 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-20
SLIDE 20

39

a5

  • - : cannot finish in 24 hours, or router crashes

10 0.100 24408652 63597 22189684 194 1713562 HSR 9 0.100 24527686 52503 22297896 14 7714 Simple Router 8 0.100 32937050 85057 29942773 12 5964 IGOR 7 0.010 16715394 5580 16557585 AMGR 5

  • 0.011

15807146 3881 15989791 NCTU 6 0.019 15960349 6584 15661544 FGR 4

  • 0.053

14854998 1889 15685923 BoxRouter 2

  • 0.100

14477335 284 16085928 FastRoute 3.0 3

  • 0.086

14552331 1058 15929290 NTUgr 1

  • 0.088

14186746 1030 15555389 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 40

b1

  • - : cannot finish in 24 hours, or router crashes

10 0.041 8448900 5432 8116300 168 866118 HSR 9 0.030 8025354 4494 7791304 14 14092 Simple Router 8 0.100 11945459 45795 10859508 6 2232 IGOR 5

  • 0.049

5851493 1141 6153326 AMGR 7 0.053 6473331 6704 6146808 NCTU 6 0.026 5882474 4194 5733109 FGR 4

  • 0.049

5420102 1147 5698086 BoxRouter 2

  • 0.100

5247776 232 5830862 FastRoute 3.0 3

  • 0.067

5398117 839 5784460 NTUgr 1

  • 0.088

5138273 586 5631048 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-21
SLIDE 21

41

b2

  • - : cannot finish in 24 hours, or router crashes

9 0.100 15449036 85484 14044578 104 1207512 HSR 10

  • Simple Router

8 0.065 16718933 44227 15695449 6 2680 IGOR 4

  • 0.004

9605806 13287 9646199 AMGR 6 0.027 9250349 22927 9004580 2 24 NCTU 3 0.000 9142997 14287 9142997 FGR 1

  • 0.100

8138447 2346 9042719 BoxRouter 7

  • 0.100

8838157 673 9820174 4 142 FastRoute 3.0 5 0.006 9777096 15862 9718446 NTUgr 2

  • 0.100

8153192 594 9059102 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 42

b3

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.100 31819946 85113 28927224 8 1706 IGOR 7 0.047 14352683 6343 13710765 AMGR 6 0.078 14238785 10955 13204201 NCTU 5 0.036 13675523 5256 13200687 FGR 2

  • 0.100

11819801 380 13133112 BoxRouter 3

  • 0.100

11848929 187 13165477 FastRoute 3.0 4

  • 0.100

12215943 296 13573270 NTUgr 1

  • 0.100

11767771 259 13075301 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-22
SLIDE 22

43

b4

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.041 45174003 85091 43403139 26 10102 IGOR 7 0.031 25476181 72421 24698274 4 656 AMGR 4

  • 0.017

22312076 31275 22697006 2 364 NCTU 5 0.041 24114597 85513 23162928 4 414 FGR 6 0.013 23459407 52644 23156271 4 472 BoxRouter 3

  • 0.100

21909974 2480 24344416 2 206 FastRoute 3.0 2

  • 0.030

23544085 24785 24281795 8 188 NTUgr 1

  • 0.099

20788825 7533 23075825 2 182 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 44

n1

  • - : cannot finish in 24 hours, or router crashes

9

  • 0.063

6661499 20416 7107317 58 258496 HSR 10

  • Simple Router

8 0.000 9409745 60539 9409745 6 2498 IGOR 7 0.010 5091039 72086 5040263 2 112 AMGR 4

  • 0.074

4253556 16675 4595493 2 30 NCTU 3 0.019 4773578 84737 4682710 2 8 FGR 5 0.012 4750677 74488 4694504 2 44 BoxRouter 6

  • 0.100

4408315 772 4898128 2 76 FastRoute 3.0 2 0.008 5070766 69021 5032685 2 6 NTUgr 1

  • 0.100

4187624 306 4652916 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-23
SLIDE 23

45

n2

  • - : cannot finish in 24 hours, or router crashes

10 0.100 12769721 16643 11608837 34 100080 HSR 8 0.100 12304622 23155 11186020 Simple Router 9 0.098 15583956 1870 14186747 IGOR 7 0.039 8367355 664 8055326 AMGR 6 0.012 7671042 419 7578450 NCTU 5

  • 0.016

7470967 259 7588737 FGR 4

  • 0.066

7096327 109 7594436 BoxRouter 1

  • 0.100

6862635 18 7625150 FastRoute 3.0 3

  • 0.100

6972768 31 7747520 NTUgr 2

  • 0.093

6863619 67 7571354 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 46

n3

  • - : cannot finish in 24 hours, or router crashes

8 0.011 14101446 85561 13944019 2180 1165162 HSR 9

  • Simple Router

9

  • IGOR

7 0.004 11436982 74803 11396687 186 37828 AMGR 6

  • 0.004

10582527 65912 10622537 206 35310 NCTU 5 0.011 10675593 85640 10555856 396 34850 FGR 4 0.009 11008981 82615 10907887 828 32404 BoxRouter 3

  • 0.100

9833414 1892 10926015 734 31650 FastRoute 3.0 1

  • 0.021

17584317 48563 17968738 432 31106 NTUgr 2

  • 0.100

9584523 7746 10649470 204 31454 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-24
SLIDE 24

47

n4

  • - : cannot finish in 24 hours, or router crashes

10 0.001 18340623 68849 18327002 242 881026 HSR 9

  • 0.054

17346202 26747 18332861 26 8260 Simple Router 8 0.013 24042662 85048 23735598 14 7742 IGOR 5 0.004 14317575 72241 14267375 2 218 AMGR 3

  • 0.009

12728614 58302 12842299 2 162 NCTU 7 0.013 13128517 85220 12959353 2 262 FGR 4 0.008 13052341 78225 12947326 4 200 BoxRouter 6

  • 0.100

12214865 596 13572072 4 226 FastRoute 3.0 1

  • 0.001

14367521 67087 14378346 2 142 NTUgr 2

  • 0.100

11690879 4023 12989866 2 152 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 48

n5

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.100 44709867 85083 40645334 22 18762 IGOR 7 0.055 25989940 15062 24640594 AMGR 6 0.057 24808532 15542 23480222 NCTU 5 0.031 24015967 9963 23295876 FGR 2

  • 0.071

21636760 1700 23294032 BoxRouter 3

  • 0.100

21707510 330 24119456 FastRoute 3.0 4

  • 0.072

22811988 1679 24577706 NTUgr 1

  • 0.100

20849787 854 23166430 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-25
SLIDE 25

49

n6

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.100 35402630 85072 32184209 16 10836 IGOR 7 0.033 19680418 6980 19049367 AMGR 6 0.025 18839950 6078 18377894 NCTU 5 0.026 18503078 6194 18030096 FGR 4

  • 0.046

17156317 1785 17975468 BoxRouter 2

  • 0.100

16793570 242 18659522 FastRoute 3.0 3

  • 0.083

17017946 935 18555976 NTUgr 1

  • 0.091

16093031 818 17695878 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router 50

n7

  • - : cannot finish in 24 hours, or router crashes

9

  • HSR

9

  • Simple Router

8 0.004 69228559 85152 68922133 14 25940 IGOR 7

  • 0.004

37655638 72933 37825580 4 3582 AMGR 6

  • 0.019

33721528 56929 34367258 2 3094 NCTU 5 0.005 35200840 86068 35023501 4 1458 FGR 2 0.004 35266586 84743 35120201 2 208 BoxRouter 4

  • 0.100

32273380 11391 35859311 6 588 FastRoute 3.0 3 0.006 37427060 86732 37222081 2 310 NTUgr 1

  • 0.100

31821609 8433 35357343 2 68 NTHU-Route rank CPU-factor CPUW-TWL CPU/s TWL MOF TOF router

slide-26
SLIDE 26

51