Swinburne University of Technology 1
O. Atwin Calchand
Internship Seminar : Projects and Progress
Centre for Advanced Internet Architectures (CAIA)
Swinburne University of Technology
18th February 2010
http://caia.swin.edu.au 18 February 2010 Page 2CAIA Seminar
Overview
� Router Project
� Why Green?
� Router Project Goals
� Experiment
� Results
� Netbook Project
� Introduction
� Goals
� Experiment
� Results
� Conclusion and Thanks
Swinburne University of Technology 2
http://caia.swin.edu.au 18 February 2010 Page 3CAIA Seminar
Why Green?
� Why Go Green?1
� Electricity generation more expensive
� PCs and monitors account for more than 40% of all ICT CO2
emissions. Data Centres account only for 23%.
� In Australia,2
� ICT sector generates 7.94 Million Tons of CO2
�Similar to the Civil Aviation and Metal production industries
� DoD reports savings of $5M per year, by just shutting down idle PCs.
1. Gartner, Conceptualising 'Green' IT and Data Center Power and Cooling Issues, Sept. 20072. Abeywickrama, Y. (2010, March). ACSView. In PCAuthority Magazine, 15.
http://caia.swin.edu.au 18 February 2010 Page 4CAIA Seminar
Router Project
� Goals of the project
� Analyse power levels
�With different components active
�With increasing amount of traffic
�At Idle and Transmit phases
� Compare power usage of different routers
� Linksys AG041 (12 W under max. load)
�Cisco 837 (18 W under max. load)
�Netgear DG834G (12 W under max. load)
� Marketed as a Green Router
Swinburne University of Technology 3
http://caia.swin.edu.au 18 February 2010 Page 5CAIA Seminar
Experiment
Python script to retrieve voltages from Oscilloscope
Data transfer between client and server using iperf with 2,10,25,50 and 100 parallel streams of traffic.
http://caia.swin.edu.au 18 February 2010 Page 6CAIA Seminar
Results
Boot Post-Boot (Idle) 1 Ethernet Port Only DSL Port Only Ethernet and DSL
Linksys AG041 6.52 6.94 7.06 7.33 7.37
Cisco 837 5.25 5.41 5.62 5.44 5.76
Netgear DG834G 3.91 4.92 5.15 5.01 5.19
0
1
2
3
4
5
6
7
8
Avergae Power (W)
Power Usage with different components active
Swinburne University of Technology 4
http://caia.swin.edu.au 18 February 2010 Page 7CAIA Seminar
Results: Data Transfer
7.214 7.211 7.211 7.211 7.210
5.82 5.79 5.82 5.89 5.89
5.15 5.12 5.11 5.11 5.11
0
1
2
3
4
5
6
7
8
2 Streams 10 Streams 25 Streams 50 Streams 100 Streams
Average Power (W)
Traffic Sent (Parallel Streams)
Power Usage Comparison Chart
Linksys AG041 Cisco 837 Netgear DG834G
http://caia.swin.edu.au 18 February 2010 Page 8CAIA Seminar
Results: Idle and Transmit Phases
7.05
7.10
7.15
7.20
7.25
7.30
7.35
7.40
7.45
1 17 33 49 65 81 97 113
129
145
161
177
193
209
225
241
257
273
289
305
321
337
353
369
385
401
417
433
449
465
481
497
513
529
545
561
577
593
609
625
641
657
673
689
705
721
737
753
769
785
801
817
833
849
865
881
897
Power (W)
Time (s)
Linksys AG041Idle & Transmit Power Levels
Idle Power (W) Mean Idle Power (W) TX Power (W) Mean TX Power (W)
Power (W) Mean
Idle 7.32
TX 7.26
Swinburne University of Technology 5
http://caia.swin.edu.au 18 February 2010 Page 9CAIA Seminar
Results: Idle and Transmit Phases
4.95
5.00
5.05
5.10
5.15
5.20
5.25
5.30
1 17 33 49 65 81 97 113
129
145
161
177
193
209
225
241
257
273
289
305
321
337
353
369
385
401
417
433
449
465
481
497
513
529
545
561
577
593
609
625
641
657
673
689
705
721
737
753
769
785
801
817
833
849
865
881
897
Power (W
)
Time (s)
Netgear DG834GIdle & Transmit Power Levels
Idle Power (W) Mean Idle Power (W) TX Power (W) Mean TX Power (W)
Power (W) Mean
Idle 5.21
TX 5.14
http://caia.swin.edu.au 18 February 2010 Page 10CAIA Seminar
Results: Cisco 837 Idle and TX Phases
5.20
5.30
5.40
5.50
5.60
5.70
5.80
5.90
6.00
6.10
1 17 33 49 65 81 97 113
129
145
161
177
193
209
225
241
257
273
289
305
321
337
353
369
385
401
417
433
449
465
481
497
513
529
545
561
577
593
609
625
641
657
673
689
705
721
737
753
769
785
801
817
833
849
865
881
897
Power (W)
Time (s)
Cisco 837Idle & Transmit Power Levels
Idle Power (W) Mean Idle Power (W) TX Power (W) Mean TX Power (W)
Power (W) Mean
Idle 5.62
TX 5.85
Swinburne University of Technology 6
http://caia.swin.edu.au 18 February 2010 Page 11CAIA Seminar
� Idle state consumes more power in two cases
� LEDs play an important role in power consumption
� 7 W Router � $11 every year
� Saving 60 mW/h for 12 h� $0.05 savings per year
� Consider the possible savings with the millions of routers around the globe
� More green products on the market
� Netgear and D-Link mainly
�No CPU stepping technology
Conclusion
http://caia.swin.edu.au 18 February 2010 Page 12CAIA Seminar
� Why Netbooks?3
� Affordable and portable
� Suitable for simple tasks such as web browsing and word processing
� Low power usage � long battery life
� Becoming mainstream devices
�Shipments increased sevenfold in Q1 2009 as compared to Q1 2008 (8% of all PC shipments)
� Acer holds 32% of netbook market
Netbook Project
3. Shah, A., 2009. Netbook shipments rise sevenfold during Q1. Network World [internet] 1 May. Retrieved February 4, 2010, from http://www.networkworld.com/news/2009/050109-netbook-shipments-rise-sevenfold-during.html
Swinburne University of Technology 7
http://caia.swin.edu.au 18 February 2010 Page 13CAIA Seminar
Netbook Project
� Acer Aspire One Pro P531h (30 W under max load)
� Intel Atom Processor N270 at 1.6 GHz (TDP 2.5 W)4
� Mobile Core 2 Duo generates TDP 35 – 65 W
� Ubuntu Netbook Remix 9.10 (Released October 2009)
� Goal: Determine the power usage of components
� BIOS, Bootloader
� Screen
� Hard Disk Drive
� Wireless
� Processor
4. Intel Corporation. (2009) . Intel Atom Processor N270. Retrieved February 12, 2010, from http://ark.intel.com/Product.aspx?id=36331
http://caia.swin.edu.au 18 February 2010 Page 14CAIA Seminar
Experiment
Swinburne University of Technology 8
http://caia.swin.edu.au 18 February 2010 Page 15CAIA Seminar
Results: BIOS and Bootloader
5.20
5.40
5.60
5.80
6.00
6.20
6.40
6.60
6.801 10 19 28 37 46 55 64 73 82 91 100
109
118
127
136
145
154
163
172
181
190
199
208
217
226
235
244
253
262
271
280
289
298
307
316
325
334
343
352
361
370
379
388
397
406
415
424
433
442
451
460
469
478
Power (W)
Time (s)
BIOS & BootLoader Power Usage
Bios Power (W) Mean Bios Power (W) BootLoader Power (W) Mean BootLoader Power (W)
Power (W) Bios BootLoader
Min 5.794 6.077
Max 6.444 6.630
Mean 6.035 6.347
Std Dev 0.099 0.083
http://caia.swin.edu.au 18 February 2010 Page 16CAIA Seminar
Results: Screen Tests
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
1 12 23 34 45 56 67 78 89 100
111
122
133
144
155
166
177
188
199
210
221
232
243
254
265
276
287
298
309
320
331
342
353
364
375
386
397
408
419
430
441
452
463
474
485
496
507
518
529
540
551
562
573
Power (W)
Time (s)
Screen Power Usage
Idle Power (W) Mean Idle Power (W)
Black Power (W) Mean Black Power (W)
White Power (W) Mean White Power (W)
MinBrightness Power (W) Mean MinBrightness Power (W)
ScreenOff Power (W) Mean ScreenOff Power (W)
Power (W) Idle Black White MinBr Off
Min 5.90 4.93 4.31 4.58 2.98
Max 6.89 6.29 9.68 5.21 3.57
Mean 6.31 5.35 5.08 4.89 3.28
Std Dev 0.12 0.16 0.25 0.09 0.07
Screen uses approx 3.13 W
Black and White reduction 5%
Swinburne University of Technology 9
http://caia.swin.edu.au 18 February 2010 Page 17CAIA Seminar
Results : Screen
Yates, D., 2009. ‘Black and White power consumption – 24 more LCD monitors tested’, Darren Yates blog, weblog post, 23 September, viewed 8 February 2010, <http://darrenyates.com.au/?p=1219>
Darren Yates is a contributing editor for the Australian PC User Magazine.
http://caia.swin.edu.au 18 February 2010 Page 18CAIA Seminar
Results: HDD
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
1 12 23 34 45 56 67 78 89 100
111
122
133
144
155
166
177
188
199
210
221
232
243
254
265
276
287
298
309
320
331
342
353
364
375
386
397
408
419
430
441
452
463
474
485
496
507
518
529
540
551
562
573
584
595
Power (W
)
Time (s)
Idle Usage with HDD and USB
HDD Power (W) Mean HDD Power (W) USB Power (W) Mean USB Power (W)
Power (W) HDD USB
Min 5.90 3.68
Max 6.89 6.51
Mean 6.31 3.97
Std Dev 0.12 0.13
HDD power approx 2.34 W
Swinburne University of Technology 10
http://caia.swin.edu.au 18 February 2010 Page 19CAIA Seminar
Results: Wireless Test
0.00
2.00
4.00
6.00
8.00
10.00
12.001 17 33 49 65 81 97 113
129
145
161
177
193
209
225
241
257
273
289
305
321
337
353
369
385
401
417
433
449
465
481
497
513
529
545
561
577
593
609
625
641
657
673
689
705
721
737
753
769
785
801
817
833
849
865
881
897
Power (W
)
Time (s)
Associated & NotAssociated Power Usage
Associated Power (W) Associated Mean Power (W) NotAssociated Power (W) NotAssociated Mean Power (W)
Power (W) Wifi Associated Wifi Not Associated
Min 3.34 4.52
Max 9.69 10.07
Mean 4.85 6.39
Std Dev 0.76 0.14
http://caia.swin.edu.au 18 February 2010 Page 20CAIA Seminar
Results: Wireless Test (New drivers)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
1 17 33 49 65 81 97 113
129
145
161
177
193
209
225
241
257
273
289
305
321
337
353
369
385
401
417
433
449
465
481
497
513
529
545
561
577
593
609
625
641
657
673
689
705
721
737
753
769
785
801
817
833
849
865
881
897
Power (W
)
Time (s)
Associated & NotAssociated Power Usage
Associated Power (W) Mean Associated Power (W) NotAssociated Power (W) Mean NotAssociated Power (W)
Power (W) Wifi Associated WiFi Not Associated
Min 6.64 4.85
Max 7.97 10.60
Mean 7.24 6.64
Std Dev 0.16 0.29
Swinburne University of Technology 11
http://caia.swin.edu.au 18 February 2010 Page 21CAIA Seminar
Results: Wireless Test (Rx & Tx)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.001 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106
113
120
127
134
141
148
155
162
169
176
183
190
197
204
211
218
225
232
239
246
253
260
267
274
281
288
295
302
309
316
323
330
337
344
351
358
365
372
379
386
393
Power (W
)
Time (s)
RX and TX Power Usage
RX Power (W) TX Power (W) Mean RX Power (W) Mean TX Power (W)
Power (W) Min Max Mean Std Dev Throughput Goodput
RX 5.55 13.07 8.95 1.12 3.2 MB/s 3.16 MB/s
TX 5.54 12.69 7.60 1.02 523 KB/s 447 KB/s
http://caia.swin.edu.au 18 February 2010 Page 22CAIA Seminar
Results: Processor
800 1070 1330 1600
Idle 6.59 6.4 6.34 6.53
1 Core 6.76 6.86 6.91 7.09
2 Cores 6.76 6.89 6.96 7.57
5.6
5.8
6
6.2
6.4
6.6
6.8
7
7.2
7.4
7.6
7.8
Average Power (W)
Clock Speed (MHz)
Idle and Loaded Cores Power Usage
Swinburne University of Technology 12
http://caia.swin.edu.au 18 February 2010 Page 23CAIA Seminar
Results: Different core speeds
Idle 1 Core 2 Cores
800 6.59 6.76 6.76
1600 6.53 7.09 7.57
800&1600 6.56 7.83 8.20
5
5.5
6
6.5
7
7.5
8
8.5
9
Average Power (W)
Idle and Loaded Cores Power UsageFirst Core at 800 MHzSecond Core at 1.6 GHz
(Green Graph)
800 1600 800&1600
http://caia.swin.edu.au 18 February 2010 Page 24CAIA Seminar
Conclusion & Future Work
� Having a dark background does not conserve power with newer LCD displays.
� LCD Display and HDD main power sinks, at 49% and 37%.
� Drivers play an important role in power usage
� Using only 1 core does not result in considerable power saving
� Future work will look at
� Power draw due to HDD read/write operations / SSDs
� Graphics card power consumption
� Comparing Windows and Linux power usage
� Looking at how much power saving can be realised through clever management of processor speed
Swinburne University of Technology 13
http://caia.swin.edu.au 18 February 2010 Page 25CAIA Seminar
Acknowledgments
� Grenville, for giving me the opportunity to work with CAIA again.
� David and Lawrence, for their help with the BART Network
� Lachlan, for his insight into my work
� Jason and Philip, for their unwavering encouragement and support.
O. Atwin Calchand
Internship Seminar : Projects and Progress
Centre for Advanced Internet Architectures (CAIA)
Swinburne University of Technology
18th February 2010
Questions ?