Battery life continuously tops the list of complaints that people have about Ultra mobile PC’s and over the last 48 hours there has been a huge amount of talk from Intel, blogs and potential customers about Intels new ultra mobile PC platforms and how battery life is going to improve. I thought it might be worth giving a little time and thought to Intel’s Ultra Mobile platform (UMP) 2007, putting some figures down and trying to work out what UMP 2007 really means in terms of battery life. Its not an easy thing to do when there are no devices available but we’ve seen some figures now and luckily, I have some real-life figures to use from testing and playing with the Q1 Ultra at CeBIT in. Read on for details about UMP 2007 and battery life.
Before I go into calculations, what we must be aware of is that Stealey, the new CPU, isn’t a major step forward in technology. Its a Pentium core built with the same 90nm production process as the previous Pentium M. At the maximum, the whole chipset and CPU combination is only 30% more power efficient and considering that the clockrate is over 10% less and the L2 cache size is reduced by 75% [datasheet] its not that much of a saving. The only significant thing that Intel have done is to implement the Enhanced Deep Sleep state which under certain conditions, can bring the power requirements down. On average though, if you were to plug the UMD CPU and chipset into the average 2006 device, you would save about 1.5 watts or about 15%. The savings would give you about another 20-25 minutes on your average 2006 UMPC. Not a great deal especially when you consider the deterioration rate of batteries. After one year you may not even notice the difference.
UMP 2007 is not about new technology, its about best-of-breed.
What Intel have done in addition to optimising their silicon is to take control of the prototyping and pull together best-of-breed components which save important percentages in areas other than the CPU and chipset. Most significantly, the screen. LED backlit screens save a great deal of energy over older CCFL backlit screens and there seem to have been advances in LED screen tech over the last 6 months too. In addition, savings can be made with the radios. Especially the WiFi radio. Some of the older WiFi modules can sap a huge 4W of power. By replacing these old components, it can be brought down to sub 1W levels. Similar savings can be made with hard drives, audio components and even by encouraging better programming practices. By controlling all these variables, using their best partners, building and owning the design right up until it is licensed out for manufacturing, they have been able to create products that bring real advantages to the user in terms of design and battery life. Potentially. We mustn’t forget that none of these products have reached the hands of customers yet.
So what’s the end effect of all this work? How does it effect battery life? Lets take the Samsung Q1 Ultra as our benchmark and see what we get. I was lucky enough to have had a good amount of time with the Q1 Ultra at CeBIT and at that time, took note of the battery figures I saw. The Q1 Ultra is a device with a 31W/hr 4-cell standard battery and at CeBIT, myself and a few others had the privilege of testing it out and giving it a heavy beating for a good 30 minutes. At the end of that, the battery life meter showed over 1.5 hours left on 45% charge. So that’s well over 3 hours on mid to heavy use which compares well with the Samsung figure of 4 hours for the basic 4-cell battery. Lets call it 3.5 hours average use. 3.5 hours from a 31w/hr battery gives a good average drain rate of just under 9Watts. It’s not a fantastic step forward but its 20-25% more efficient than most of the UMPCs out there today with the added advantage of extra screen brightness and GPU power thrown in.
What sort of battery life you would get on the smaller devices? The 4.8″ screen devices? When considering the small 4.8″ devices, there are three things to remember. One, the screen is much smaller and needs much less energy to backlight it. Two, the smaller devices will not be able to take big 30W/hr batteries. Three, there’s an option to drop in the 600Mhz Stealey CPU. All other variables stay the same assuming that the radio components, memory and HDD are similar. I estimate that the screen size reduction would save 1W under average conditions. This is based on figures I’ve seen for devices like the Raon Digital Vega. Dropping the clockrate to 600Mhz would save a proportional amount of energy which would equate to about 300mw average. In total, the smaller device power reduction would be 1300mw bringing the average drain down to about 7.5 watts. Its a very nice 17% power saving over the larger device. Dont forget though, that you are left with a 600Mhz processor in your hands. Not very 2007! In addition, you can’t just snap on a 31W/hr battery to a small device. It’s too heavy and too big. You’re restricted to something like a 20W/hr battery meaning you will end up with about 3 hours battery life. Its not bad at all really, especially if you compare this to the latest Flipstart UMPC which uses a 40W/hr battery and brings well under four hours usage to the user at an average drain of around 11watts. [Estimated from tests. Its interesting to look back on my comments here too.]
So in summary, while UMP 2007 is a good step forward (and in line with my year-on-year predictions) its not a massive leap forward in technology. Its enabled smaller devices to be built and I would expect to see 15-20% battery life improvement which pushes the 7″ devices up into the 4 hours bracket. If I’m right with those figures, it certainly doesn’t push VIA out of the game. The C7-M is still a more efficient processor than Stealey and now that VIA have a new GPU (VN896) they could do much the same as what Intel have just done. In fact, the Q1b already returns 4 hours of battery life and is a good example of what can be done. There’s also the ‘John’ system on chip design. What UMP 2007 does in 1700mm2, John will do in 1250mm2 [Read this for more info] All it takes is the right components and a strict control over the design and manufacturing process. Computex 2007 is coming up in June and VIA have a chance to show what is possible with their platform. If they pull together the C7-M and a DX9 capable GPU into their CoreFusion product and get it to market with some nice screen and radio technology along with their usual price advantage you could see devices beating UMP 2007 for efficiency and coming in at a cheaper price.
What UMP 2007 really brings is not battery life or smaller die sizes or even a high quality ecosystem. It brings proof that Intel is prepared to put a lot of money behind UMPCs. The market research is over for Intel and they are now, as they said yesterday, investing ahead of the curve. If VIA want to play in the consumer mobile x86 game, they need to work very hard.
[Figures here are based on my own knowledge, research and experience.]
