NRDC and CLASP just released a new study on the energy use of computer graphics card. The study proposes a new approach to settings appropriate energy allowances for graphics cards in computer energy efficiency standards and labeling programs. It tested a representative sample of recent graphics cards and computers and provide recommended allowance levels for computer programs.
The study's results provide an objective benchmark to validate graphics allowances currently proposed by the draft ENERGY STAR v6.0 specification and Ecodesign Computer measure in the EU. The comparison of draft allowances with the study's results shows that both standards are currently proposing significantly higher allowances than necessary in high-end graphics categories, creating a loophole in the programs where computers equiped with the latest graphics cards will get a large unwarranted credit to qualify, potentially allowing inefficient computers to qualify.
In the age of computer gaming, video editing, computer-aided design and ultra-high resolutions, discrete (aka add-in) graphics cards are the engines of today’s media oriented computers. They can also be responsible for a large share, and in some cases the majority, of the energy used by a computer. There are over a billion computers in use in the world today, consuming the equivalent output of 90 medium-sized 500 MW coal-fired power plants, or six times the electricity used annually by all the households in New York City!
This has not gone unnoticed by policy makers worldwide: it is much more cost-effective and less polluting to increase the energy efficiency of computers than to build new power plants. Several governments around the world have recently adopted computer efficiency standards, such as India and Australia, or are developing some such as the European Union, China and California. Computers have also been covered by the US EPA ENERGY STAR program for over a decade.
Setting appropriate energy allowances for discrete graphics cards is becoming increasingly critical to the effectiveness of computer efficiency standards: our testing found that discrete graphics cards can represent between 20 and 60 percent of the entire computer’s energy consumption.
To address this issue, the Natural Resources Defense Council (NRDC) and the Collaborative Labeling & Appliance Standards Program (CLASP), in collaboration with Ecova, launched a graphics card testing study. The goal of the study was to provide policy makers with objective and transparent measurements of discrete graphics card energy requirements, in support of setting effective efficiency standards and labeling programs.
The study developed a test methodology that measured the power difference between a computer with a discrete graphics card and the same computer without the card (using integrated graphics). Both measurements were performed with the computer in idle mode, i.e. not performing any graphics intensive task. Idle mode was used as a proxy for active mode consistently with the ENERGY STAR test method.
Key findings from the study include:
- The power impact of each discrete graphics card varied significantly from computer to computer, validating that a number of system-specific factors other than the card itself impact system power demand when a discrete graphics card is installed. For example, the integrated graphics processing unit can be switched off, and the discrete graphics card generates activity in other components;
- Power demand in idle mode generally increased as cards became more powerful, even though measurements were made in idle more when little or no graphics processing was taking place; There were large differences between cards, indicating the potential for significant energy efficiency improvements in some cards; and
- One of the two high-performance cards tested showed dramatically lower energy use in idle mode than the other one (one fifth). This card, the AMD Radeon HD 7970, was the first card on the market to feature AMD’s “ZeroCore Power” technology, which radically reduces idle power. The other major graphics chip manufacturer, Nvidia, has also released similar technology since then. This is very promising for the energy efficiency of computers using discrete graphics; however, efficiency specifications need to factor in this new technology, otherwise this could become a loophole where computers that use these cards are granted a large unwarranted allowance to qualify, even when they might not have qualified without the card.
Check out the study’s report for the detailed methodology, findings and recommendations.
NRDC’s and CLASP’s study provides policy makers with a unique dataset to set appropriate allowances for graphics cards in computer efficiency programs, helping reduce energy use by computers worldwide, saving consumers money in the form of reduced electricity bills, and curbing air pollution that endangers people’s health and the planet’s climate.