One of the basic issues of green computing is the electric energy consumption. It is especially important due to the emergence of huge data centers that operate on data stored in the cloud computing.
This topic consists of the following sections:
The work \(W\) for AC powered devices is calculated as follows:
\[
W=U_{RMS} \cdot I_{RMS} \cdot t
\]
where
Power is a work done in a unit of time: \[ P = {{W} \over {t}} \] where
Laptops
The average power consumption can range from 20W for a netbook to 110W for a gaming laptop:
model | power consumption watts |
---|---|
netbook Intel Atom N450, 1GB RAM | 20 |
laptop Intel Core i7-720QM, 4GB DDR3 RAM, 16" LED | 50 |
laptop Intel Core i3-330M, 3GB DDR3 RAM, 15.6" LED | 65 |
laptop Intel Core i7-720QM, 8GB DDR3 RAM, 17,3" Full HD | 103 |
Monitors
For some applications, especially for the artists, CRT monitors are better for the close to realistic representation of colors. For many other uses TFT surpass CRTs.
TFT monitor consumes much less power than CRT monitor. The following table compares the power consumption of some selected monitors.
type | model | Brightness 100% | Brightness 50% | Brightness 10% |
---|---|---|---|---|
TFT | Acer E225HQL 22" | 18 | 13 | 9 |
TFT | Philips 190BW9 19" | 31 | 21 | 12 |
TFT | BenQ FP937S 19" | 31 | 24 | 21 |
TFT | Samsung SyncMaster 204B 20" | 34 | 25 | 18 |
TFT | Samsung SyncMaster 245B plus 24" | 66 | 44 | 26 |
CRT | Iiyama Vision Master Pro 454 19" | 102 | 98 | 91 |
CRT | Sony CPD-G420 19" | 111 | 103 | 99 |
Desktops
The following table presents power consumption of selected desktop computers.
Configuration | Power [W] |
---|---|
Intel Pentium Dual-Core E5200, 2GB DDR3 RAM, SATA HD 5400 RPM, GPU Intel GMA X4500 | 110 |
Intel Core i3-530, 4GB DDR3 RAM, SATA HD 7200 RPM, GPU ATI Radeon HD 5670 | 200 |
Intel Core i7 920, 6GB DDR3 RAM, SATA HD 7200 RPM, GPU GeForce GTX 580 | 500 |
Supercomputers
The Top 500 contains the list of 500 supercomputers sorted by performance measured in FLOPS i.e. floating point instructions per second (MFLOPS is equal to \(10^6\) FLOPS, GFLOPS is equal to \(10^9\) FLOPS, TFLOPS is equal to \(10^{12}\) FLOPS).
The following table contains the top 10 supercomputers with maximal performance (Rmax). The last column shows the power consumption in kilowatts.
Rank | System | Cores | Rmax [TFLOPS] |
Power [KW] |
---|---|---|---|---|
1 | Tianhe-2 - TH-IVB-FEB Cluster | 3,120,000 | 33,862.7 | 17,808 |
2 | Titan - Cray XK7 | 560,640 | 17,590.0 | 8,209 |
3 | Sequoia - BlueGene/Q | 1,572,864 | 17,173.2 | 7,890 |
4 | K computer | 705,024 | 10,510.0 | 12,660 |
5 | Mira - BlueGene/Q | 786,432 | 8,586.6 | 3,945 |
6 | Piz Daint - Cray XC30 | 115,984 | 6,271.0 | 2,325 |
7 | Stampede - PowerEdge C8220 | 462,462 | 5,168.1 | 4,510 |
8 | JUQUEEN - BlueGene/Q | 458,752 | 5,008.9 | 2,301 |
9 | Vulcan - BlueGene/Q | 393,216 | 4,293.3 | 1,972 |
10 | Cray CS-Storm | 72,800 | 3,577.0 | 1,499 |
The important criterion for green computing is the performance in MFLOPS per watt of power consumption. For instance the fastest supercomputer Tianhe-2 has 1,901.54 MFLOPS/W.
Cloud computing
Report How Clean is Your Cloud? by Greenpeace : 1 presents the analysis of power consumption of the largest clouds (Facebook, Amazon, Apple, Microsoft, Google, Yahoo and others).
The following figure shows the electric consumption in 2007 (data from : 1 ):
Fig. 3/1: 2007 electricity consumption
The cloud computing in 2007 consumed much more power than the fifth country in the world (e.g. India).
The formula for cost of energy is as follows: \[ C={{P} \over {1000}} \cdot t \cdot p \] where
The simple calculator of energy costs is available on the page energyusecalculator.com .
Greenpeace: How Clean is Your Cloud?
P. Schmid, A. Roos: Display power consumption: CRTs Versus TFT-LCDs
Tom's Hardware, August 2010. http://www.tomshardware.com/
J. Klimkowicz: Ile kosztuje dzień pracy komputera?
PC Skan, December 2010 (in polish) http://pcskan.gadzetomania.pl/2010/12/14/ile-kosztuje-dzien-pracy-komputera
www: The Top 500 supercomputers
November 2014 http://top500.org/lists/2014/11/
www: The Green 500 supercomputers
November 2014 http://www.green500.org/lists/green201411