Ever wonder how device makers come up with their battery numbers?
Most people believe battery life for their laptop and mobile devices is always and every time just plain less than what they want, or that they deserve.
For example, Apple’s recent suggestion that iPhone 3GS users can improve battery life by turning off features was not well received by users. An LA Times reporter dubbed the battery in the new phone “iDrain.”
Why is there such a chasm between between battery life test results from vendors and our expectations?
The figures for laptop battery longevity are often based on an industry “standard” called the MobileMark 2007, a benchmark created by an outfit called BAPco, which is a group of computer makers, chip vendors and other companies, according to Newsweek Tech Editor Daniel Lyons.
Lyons‘ story is about how one BAPco member, chipmaker AMD, is urging a new or at least revised test, partly he implies because AMD chips don’t do well based on the MM07 suite compared to arch-rival Intel.
Like all tests, the MM07, has to make certain assumptions to create a uniform test environment at least insofar as what to test. But the thrust of Lyons’ piece seems to be that the tests are essentially bogus, designed from the outset to give a veneer of respectability inflated claims.
Skepticism is, or should be, inherent in journalism not to mention capitalism (caveat emptor and all that). But there’s a line between skepticism and cynicism.
My understanding is that any benchmark, even one that attempts to emulate a real-world use scenario, is always somewhat artificial. But a real benchmark is a real attempt to create a genuine repeatable comparison between different systems.
According to Lyons, “Laptops score big numbers because they’re tested with screens dimmed to 20 to 30 percent of full brightness, the Wi-Fi turned off and the main processor chip running at 7.5 percent of capacity.”
I checked the online BAPco whitepaper on MM07, which obviously is not the actual full test itself, and found some of Lyons’ assertions here to be accurate.
“Wireless” is “turned off at front panel,” according to the settings for the reference laptop (an Intel-powered Lenovo T60) on the document’s last page. Screen brightness “should be measured for a white screen while on battery and be set at the lowest possible setting, no lower than 60 nits.” According to BAPco, the “60 nit level was chosen as a reasonable ‘middle’ setting, bright enough for the display to be viewed in a common office environment, but not too bright for a darkened room.” In other words, they made some assumptions regarding how the display might be set by a user wanting to optimize battery life.
With regard to running the CPU at 7.5% of capacity, or any other capacity, the whitepaper had nothing to say, based on my quick read.
The MM07 test measures battery life in minutes for three test modules: playing a DVD, reading a document, and a series of “productivity” tasks that a user might do, such as opening a document and searching it for a keyword, creating and sending email, working with a spreadsheet and so on. These tasks run individually until the battery is exhausted.
There is a separate “performance” metric that measures the “average response time” for the tested laptop relative to the Lenovo reference system, with a performance rating of 200. This rating is intended as a “qualification” of the measured minutes in the test modules. “A rating of 100 indicates that the tested system took twice as long, on average, to perform tasks as the reference system. A rating of 400 indicates that the tested system took half the time, on average, to perform tasks as the reference system,” according to the BAPco whitepaper.
Based on my reading, BAPco’s use cases don’t include things like watching YouTube via a Wi-Fi connection on your laptop, posting photos on FlickR, being plugged into push email all the time, or running an online collaboration application. In other words, it tests the “system” narrowly understood, but not the use of that system in an increasingly networked and online world.
That may account for the big discrepancies users experience between stated battery life and their actual usage. And it certainly suggests that BAPco needs to update or at least expand the kinds of things tested by MobileMark.
You can’t do much to change your corporate laptop, or improve its internal power efficiency. But do you have specific “best practices” to optimize battery life for your laptop or handheld?
[Update: I had asked Apple if they could provide their Battery Guru to talk about mobile battery issues in general. They declined, and referred me to their online battery test results page.]
