CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf ·...
Transcript of CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf ·...
![Page 1: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/1.jpg)
CS848Advanced Topics in DatabasesDatabase Systems on Modern
HardwareSpring 2015
Ken Salem
David R. Cheriton School of Computer ScienceUniversity of Waterloo
![Page 2: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/2.jpg)
US Data Center Energy Consumption
from Growth in Data center electricity use 2005 to 2010.,Jonathan Koomey. Analytics Press, Oakland, CA. 2011http://www.analyticspress.com/datacenters.html
� cooling anddistributiondoubleconsumption
� 1% reduction≈ 1 billionkWh/year
� last year, mycondo ≈ 4500kWh
� 1% reduction≈ 200,000condos
![Page 3: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/3.jpg)
US Data Center Energy Consumption
from Growth in Data center electricity use 2005 to 2010.,Jonathan Koomey. Analytics Press, Oakland, CA. 2011http://www.analyticspress.com/datacenters.html
� cooling anddistributiondoubleconsumption
� 1% reduction≈ 1 billionkWh/year
� last year, mycondo ≈ 4500kWh
� 1% reduction≈ 200,000condos
![Page 4: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/4.jpg)
US Data Center Energy Consumption
from Growth in Data center electricity use 2005 to 2010.,Jonathan Koomey. Analytics Press, Oakland, CA. 2011http://www.analyticspress.com/datacenters.html
� cooling anddistributiondoubleconsumption
� 1% reduction≈ 1 billionkWh/year
� last year, mycondo ≈ 4500kWh
� 1% reduction≈ 200,000condos
![Page 5: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/5.jpg)
US Data Center Energy Consumption
from Growth in Data center electricity use 2005 to 2010.,Jonathan Koomey. Analytics Press, Oakland, CA. 2011http://www.analyticspress.com/datacenters.html
� cooling anddistributiondoubleconsumption
� 1% reduction≈ 1 billionkWh/year
� last year, mycondo ≈ 4500kWh
� 1% reduction≈ 200,000condos
![Page 6: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/6.jpg)
US Data Center Energy Consumption
from Growth in Data center electricity use 2005 to 2010.,Jonathan Koomey. Analytics Press, Oakland, CA. 2011http://www.analyticspress.com/datacenters.html
� cooling anddistributiondoubleconsumption
� 1% reduction≈ 1 billionkWh/year
� last year, mycondo ≈ 4500kWh
� 1% reduction≈ 200,000condos
![Page 7: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/7.jpg)
Energy Efficiency of Cloud Computing
from E. Masanet et al, The Energy Efficiency Potential of Cloud-BasedSoftware: A U.S. Case Study,
Lawrence Berkeley National Laboratory, June 2013http:
//crd.lbl.gov/assets/pubs_presos/ACS/cloud_efficiency_study.pdf
� most hosting stillin small serverrooms/closets
� step 1: move tocloud
� step 2: optimizecloud
![Page 8: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/8.jpg)
Energy Efficiency of Cloud Computing
from E. Masanet et al, The Energy Efficiency Potential of Cloud-BasedSoftware: A U.S. Case Study,
Lawrence Berkeley National Laboratory, June 2013http:
//crd.lbl.gov/assets/pubs_presos/ACS/cloud_efficiency_study.pdf
� most hosting stillin small serverrooms/closets
� step 1: move tocloud
� step 2: optimizecloud
![Page 9: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/9.jpg)
Energy Efficiency of Cloud Computing
from E. Masanet et al, The Energy Efficiency Potential of Cloud-BasedSoftware: A U.S. Case Study,
Lawrence Berkeley National Laboratory, June 2013http:
//crd.lbl.gov/assets/pubs_presos/ACS/cloud_efficiency_study.pdf
� most hosting stillin small serverrooms/closets
� step 1: move tocloud
� step 2: optimizecloud
![Page 10: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/10.jpg)
Energy Efficiency of Cloud Computing
from E. Masanet et al, The Energy Efficiency Potential of Cloud-BasedSoftware: A U.S. Case Study,
Lawrence Berkeley National Laboratory, June 2013http:
//crd.lbl.gov/assets/pubs_presos/ACS/cloud_efficiency_study.pdf
� most hosting stillin small serverrooms/closets
� step 1: move tocloud
� step 2: optimizecloud
![Page 11: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/11.jpg)
Cost of Large Data Centers
from James Hamilton, Cost of Power in Large-Scale Data Centers,Perspectives blog, Nov 2008
http://perspectives.mvdirona.com/2008/11/cost-of-power-in-large-scale-data-centers/
� “fully burdened cost of power” ≈ 42%� (+) server costs decreasing, power cost
increasing� (-) server power efficiency improving
![Page 12: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/12.jpg)
Cost of Large Data Centers
from James Hamilton, Cost of Power in Large-Scale Data Centers,Perspectives blog, Nov 2008
http://perspectives.mvdirona.com/2008/11/cost-of-power-in-large-scale-data-centers/
� “fully burdened cost of power” ≈ 42%
� (+) server costs decreasing, power costincreasing
� (-) server power efficiency improving
![Page 13: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/13.jpg)
Cost of Large Data Centers
from James Hamilton, Cost of Power in Large-Scale Data Centers,Perspectives blog, Nov 2008
http://perspectives.mvdirona.com/2008/11/cost-of-power-in-large-scale-data-centers/
� “fully burdened cost of power” ≈ 42%� (+) server costs decreasing, power cost
increasing
� (-) server power efficiency improving
![Page 14: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/14.jpg)
Cost of Large Data Centers
from James Hamilton, Cost of Power in Large-Scale Data Centers,Perspectives blog, Nov 2008
http://perspectives.mvdirona.com/2008/11/cost-of-power-in-large-scale-data-centers/
� “fully burdened cost of power” ≈ 42%� (+) server costs decreasing, power cost
increasing� (-) server power efficiency improving
![Page 15: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/15.jpg)
Data Center Server Utilization
1.00
0.01
0.005
0.02
0.03
0.015
0.025
CPU utilization
Fra
cti
on
of
tim
e
0.90.80.70.60.50.40.30.20.10
from Barroso and Hölzle, The Case for Energy-Proportional Computing,IEEE Computer 40(12), Dec 2007, pp. 33-37
� utilization of> 5000 Googleservers over 6months
� most servers10%-50%
� full idle unlikely
![Page 16: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/16.jpg)
Data Center Server Utilization
1.00
0.01
0.005
0.02
0.03
0.015
0.025
CPU utilization
Fra
cti
on
of
tim
e
0.90.80.70.60.50.40.30.20.10
from Barroso and Hölzle, The Case for Energy-Proportional Computing,IEEE Computer 40(12), Dec 2007, pp. 33-37
� utilization of> 5000 Googleservers over 6months
� most servers10%-50%
� full idle unlikely
![Page 17: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/17.jpg)
Data Center Server Utilization
1.00
0.01
0.005
0.02
0.03
0.015
0.025
CPU utilization
Fra
cti
on
of
tim
e
0.90.80.70.60.50.40.30.20.10
from Barroso and Hölzle, The Case for Energy-Proportional Computing,IEEE Computer 40(12), Dec 2007, pp. 33-37
� utilization of> 5000 Googleservers over 6months
� most servers10%-50%
� full idle unlikely
![Page 18: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/18.jpg)
Power Proportionality� energy consumption proportional to work done
� SPECpower_ssj2008 benchmark
� power range improving over time?
Dell PowerEdge R630(April 2015)
dual proc/36 cores/64 GB
Dell PowerEdge 2950 III(Dec 2007)
dual proc/8 cores/16 GB
![Page 19: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/19.jpg)
Power Proportionality� energy consumption proportional to work done� SPECpower_ssj2008 benchmark
� power range improving over time?
Dell PowerEdge R630(April 2015)
dual proc/36 cores/64 GB
Dell PowerEdge 2950 III(Dec 2007)
dual proc/8 cores/16 GB
![Page 20: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/20.jpg)
Power Proportionality� energy consumption proportional to work done� SPECpower_ssj2008 benchmark
� power range improving over time?
Dell PowerEdge R630(April 2015)
dual proc/36 cores/64 GB
Dell PowerEdge 2950 III(Dec 2007)
dual proc/8 cores/16 GB
![Page 21: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/21.jpg)
Idle-to-Peak Trend
0
0.2
0.4
0.6
0.8
1
Jul2007
Jan2008
Jul2008
Jan2009
Jul2009
Jan2010
Jul2010
Jan2011
Jul2011
Jan2012
IPR
(id
le-t
o-p
ea
k p
ow
er
ratio
)
Hardware release date
Idle-to-Peak power ratio (IPR) for various computing systems
linear average projectionIPR-2
from Varsamopoulos and GuptaEnergy Proportionality and the Future: Metrics and Directions,
In Proc. Int’l Conf. on Parallel Processing Workshops, 2010
![Page 22: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/22.jpg)
Techniques for Energy Efficiency
� dynamic server (de)provisioning� adjust number of active servers to load� idle or power down unused servers
� frequency and voltage scaling� adjust CPU frequency based on workload� lower frequency ⇒ less power consumed
� energy-aware scheduling� choose energy-efficient platform for each
workload
![Page 23: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/23.jpg)
Voltage and Frequency Scaling
System power consumption vs. TPC-C throughputin various p-states
Shore-MT, in-memory database
![Page 24: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/24.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 25: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/25.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 26: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/26.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S
� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 27: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/27.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 28: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/28.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 29: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/29.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 30: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/30.jpg)
CPU ScalingS = process feature size ratio, e.g,
32 nm to 22 nm gives S = 32/22 ≈ 1.4
Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1/S2
� ⇒ ∆ Power∝ ∆QFCV2 = 1
� ⇒ ∆ Utilization ∝ 1
post-Dennard scaling
� ∆ Quantity ∝ S2
� ∆ Frequency ∝ S
� ∆ Capacitance ∝ 1/S� ∆ Voltage ∝ 1� ⇒ ∆ Power∝ ∆QFCV2 = S2
� ⇒ ∆ Utilization ∝ 1/S2
Source: M.B. Taylor, A Landscape of the New Dark Silicon Design Regime. IEEE Micro 33(5), Aug.2013, pp. 8-19.
![Page 31: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/31.jpg)
Dark Silicon
� silicon that is not used all the time, or not usedat its full frequency
� fixed power envelope limits growth in Q or F orboth
� Denard: QF grows by S3
� post-Denard: QF grows by only S
![Page 32: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/32.jpg)
Dark Silicon Example
4 cores at 1.8 GHz
4 cores at 2×1.8 GHz
(12 cores dark)
2×4 cores at 1.8 GHz
(8 cores dark, 8 dim)
(Industry’s choice)
75% dark after two generations;
93% dark after four generations
65 nm 32 nm
Spectrum of trade-offs
between no. of cores and
frequency
Example:
65 nm → 32 nm (S = 2)
....
....
....
Source: M.B. Taylor,A Landscape of the New Dark Silicon Design Regime.
IEEE Micro 33(5), Aug. 2013, pp. 8-19.
![Page 33: CS848 Advanced Topics in Databases Database Systems on ...kmsalem/courses/cs848S15/overview.pdf · CS848 Advanced Topics in Databases Database Systems on Modern Hardware Spring 2015](https://reader030.fdocuments.net/reader030/viewer/2022040513/5e6851d55431164b5f6fe84f/html5/thumbnails/33.jpg)
Responses to Dark Silicon
� smaller chips� “dim” silicon
� reduce clock rate, or� use more space for low-power functions, e.g.,
cache,� power only part of the time
� functional specialization� fast or efficient co-processors� execution hops around