ARM Community: Why Battery Life Benchmarks Must Change for Mobile Devices - ARM Community

The current battle in the Intel vs AMD PC war has been highlighted by recent discussions on the relevance of MobileMark2007 to the consumer’s true experience in mobile computing. AMD wants to include 3DMark2006 as another piece of the puzzle (presumable it will help their position) and Intel seems to benefit from the amount of time MobileMark2007 spends in CPU idle. In reality, future changes in this area needs to go well beyond being another battle in the PC wars and instead incorporates the breadth of new smart mobile devices that are coming to market.

The emergence of netbooks, smartbooks*, and other devices that enable consumers to consume, create, and communicate demonstrate that usage models have moved beyond the one size fits all model of the PC. Consumers now have choice to buy different devices depending on their needs and lifestyles.

With this range of new devices, there are new operating systems even beyond the Linux flavors that HP, Dell, Asus, and others are shipping. Nvidia has shown some products based on WinCE, there were numerous devices based on Android shown at Computex in June, and now Google has announced the ChromeOS to be shipping in 2010. Benchmarks cannot be limited to Windows XP and Vista operating systems if consumers are going to be able to take advantage of the new choices that are being made available.

Add into the mix the fact that the majority of the new devices being made available to consumers for mobile computing are NOT using an x86 architecture that power the “traditional” PCs today. Many are System on Chips (SOCs) based on an ARM processor core with dedicated circuitry for graphics, video, communications, etc.

Finally, and most important, there is no longer a one size fits all consumer. Many consumers are now much more aware of what their individual usage patterns are. They are making buying decisions based on what suits them best based on their needs. For some it is about communicating with Facebook, Twitter, and Skype. Others are doing generic web browsing and e-mail and still others it is about multimedia including online vide such as YouTube or Hulu.

Key requirements for benchmarks moving forward are to:

- Be tied to specific usage models. The mobile phone market historically did exactly that with talk time and standby time. Consumers could mix and match how much they would talk over the days to figure out what worked best for them. New mobile computing devices just have a few more usages. Online video, locally stored video, audio playback, web browsing, pictures, others? The consumer can then mix and match what is most important to them[/indent] - Be transportable across different operating systems. Given the number of options is growing this is extremely important. Maybe this is simply a web based approach combined with the default local applications that ship with the OS (browser, media player, productivity etc).

- Be transportable across chipsets and architectures. The PC wars have created benchmarks that often appear to be tuned to one particular architecture or chipset. The beauty of the changing computing model is that diversity and innovation are creating new solutions each of which has measureable advantages and disadvantages. Some will have better graphics, others better video, and still others better communications. Benchmarks should highlight each of these based on the usage model and allow the consumer to draw conclusions based on clear information.

- Be reproducible with well documented assumptions BUT make those assumptions realistic. 20% backlight is not realistic for most users and thus it will artificially result in higher numbers than users will see. Similarly, stating browsing times while the wireless is OFF is ridiculous.

As mobile computing devices continue to evolve from just PCs to smartbooks*, netbooks, Smartphones and all the devices in between, benchmarks need to reflect this diversity and provide meaningful information for consumers to use in their decision making process. Now is the time to act and help consumers make smart choices.

What do you think? How else can the industry help consumer choice? Comment below.


Jeff Chu, Segment Marketing Manager, Mobile Computing, ARM, has been on the forefront of ultra mobile computing for over 8 years. From webpads to smart displays to UMPC there has always been the criticism that it's not a real PC. Well, now the computing world is flipping and low power mobile gadgets are becoming the preferred way to consume content and interact with the web. To quote Victor Hugo “There is nothing more powerful than an idea whose time has come.”

Shortlink to this post: http://bit.ly/14BHuM yz76gsne4t

* In the territory of the Federal Republic of Germany, the use of the term "Smartbook" in connection with portable computers is reserved exclusively to Smartbook AG, Germany.