The number of bugs in a chip is relatively proportional to the number of transistors - Bob Colwell, former Intel chief architect
THE SLOW PACE of battery development is what ultimately will hold up innovation in mobile devices.
In recent years smartphones and tablets have become the most talked about mobile computing devices, but they are just the latest and perhaps most visible containers for stagnant battery technology. One can only wonder how much laptop development stagnated due to the slow pace of battery innovation.
Chip firms have spent years and billions tackling the battery life problem from a brute force point of view through physically shrinking transistors and engineering higher performance. The industry hoped that increasing top-line performance and throttling cores and chip frequencies would help increase battery life, and while that works in theory, in practice people wanted smaller, thinner devices that had lower capacity batteries.
Those chip firms that have thrown money at manufacturing have always been keen to say it gives them a business advantage. AMD's founder Jerry Sanders famously said, "Only real men have fabs" and several senior Intel executives have said that companies without chip fabrication plants simply won't be able to compete in the future. Looking at how AMD, Nvidia and Qualcomm have struggled with TSMC's 28nm process node issues, it would seem that Intel has a point, but higher chip performance achieved through process node shrinkage will not deliver the mobile devices that consumers want. The fact is that processor performance growth has outstripped battery performance by several orders of magnitude.
When Nvidia launched its Tegra 2 system-on-chip with the help of LG's Optimus 2X, its CEO Jen-Hsun Huang told the audience that having multiple cores allows each core to run on lower voltages for more of the time. Huang's comments have been echoed by many in the semiconductor industry, and since voltage and power consumption are linked exponentially it is an argument that in theory holds water.
In practice however, things haven't changed for smartphone users. Despite the number of cores increasing from one to four - or in the case of the Tegra 3, five - the vast majority of smartphones still need charging after a day's use. The truth is that processors have little affect on the real-world battery life of devices such as smartphones and tablets, and that is why, when Intel finally got into the smartphone game with what many thought was a power-hungry architecture, reviewers found that the device's battery life was on a par with that of its ARM-based contemporaries.
Smartphones are not the only devices where battery life hasn't improved despite the internal hardware having seen considerable development. When solid-state drives first arrived, the talk was of longer laptop battery life as the spinning platter was ditched for the NAND flash chip, but once again that hasn't happened.
So while semiconductor makers spend vast sums of money on trying to improve battery life, something that many smartphone users would take over yet another CPU core, battery manufacturers still can't get their innovations to market. As consumers show a desire for thinner devices that require higher density batteries, it doesn't look like users will see an increase in battery life for some time, and this requires chip vendors to look at how they spend research funds.
From a distance it is painfully obvious that chasing Moore's Law is an exercise in futility. Intel has hung onto it by ditching the race for MHz in favour of multi-core architectures, but the truth is that doubling the transistor count every 18 months is fast becoming pointless from a user's perspective.
Intel in particular won't drop Moore's Law, not because its most famous employee coined the term but that by giving up on it now it is likely to be seen as a failure for the company. The problem is that even Intel with its $40bn in cash can't keep bankrolling $10bn process node shrinks if the useful performance gain isn't large enough to justify users going out and buying new devices that have Intel chips.
As mobile devices represent the biggest sales growth opportunity for chip vendors, instead of investing billions and a decade of effort into shrinking transistors, chip makers should look to sink a chunk of that research money into battery development. Otherwise chip vendors and device makers might soon find that playing games won't help flog their new processors and devices, just as they have with laptops.
Ultimately what chip vendors need to do is drop the brute force approach to reducing power demands and work more closely with the battery industry, if they really want to see their billions in research and development spending exploited effectively in mobile devices. µ