Intel is very serious about low power chips, although it won’t have them until 2013. The company showed off the long-rumored Haswell chips at its developer forum on Tuesday, which it says can run all day and offer a 20x reduction in power compared with existing chips. Intel also convinced Google to support x86 chips for its Android tablet and phone software. So Intel is serious about mobile, and enabling mobile devices with long battery life, but will the industry buy it?
That question won’t be answered today, but Intel is in the very least trying to avoid being a mobile loser as Qualcomm and other vendors using the ARM architecture make strides inside tablets and smartphones. Intel’s fighting to control the consumer computing market as consumers want low-power portable devices, while also trying to continue its expansion on the server side, where it has seen tremendous growth.
The Haswell products, which unfortunately won’t be out until 2013, will have 10 days of standby battery life, which puts Intel into the same league as ARM’s designs. However, it’s unclear where vendors such as Nvidia, Qualcomm, Samsung, Apple and other ARM licensees might be in terms of performance and power consumption at that time. And while software platforms are now tuned to Intel and ARM-based chips, Intel lost out on a huge advantage by being slow to cut power consumption. A year ago, before Microsoft committed to support ARM-based chips, Intel had a significant advantage.
Intel is also going to have to fight to get its way into handsets, which can take a long time. For example, Nvidia launched its first application processor in 2008 and only scored some major wins in devices in 2011. Handset makers aren’t eager to pick up new-fangled chips in their devices, so it can take a while. However, in the tablet market, Intel could pick up traction, as enterprise customers are already in favor of using Intel on the devices judging from Cisco’s efforts with the Cius tablet.
Haswell, which Intel showed off running on a solar panel, also may have a spot in the micro server market, according to Andrew Feldman, CEO of SeaMicro. SeaMicro makes a rack of servers that use Atom chips today, but could end up using high-end Atom chips or low-end Haswell chips. By covering both ends of the low-power market, Intel is signaling it’s serious about low power, both for the client side and on the server side.
I don’t buy it. Intel is very good at making big misleading claims. Every site out there has put a headline like “Intel chips to have 20x less power consumption in 2013″ – which isn’t even what Intel said, but they let it sound like that’s what they said.
So does that mean Intel’s chips in 2013 will go from the 40W TDP in laptops today, to 2W? No, of course not. What they were talking about is the power consumption in idle mode, and it’s even in their PR image at the top saying “Connected Stand-by Power”.
So it’s all about power management when in stand-by. It’s not about power consumption when in use. which of course they couldn’t have made 20x less in just 2 years. But everyone out there took it to mean that basically Intel’s chips will have a 2W power footprint in 2013, just like ARM chips.
At best, they might be able to do that with Atom chips, which now range from 5W a single core Atom, to 10W a dual core Atom. But my guess is they will only be able to reach a 1-2W TDP with a single core Atom by 2013.
Well, hate to break it to Intel, but in 2013 we might even see 8 core 2.5 Ghz ARM chips, that not only will have the same low TDP as always, but will also cost as little as the high-end ARM chips today, which is to say, up to 5x less than an Atom chip.
So Intel still has their work cut out for them. I think Google’s decision to support Intel was mostly political, not done on technical merits. They’ve been partnering with Intel on Google TV and Chromebooks and selling them at prices higher than they should’ve been, Paul Otellini is also on their board, and now with Qualcomm and Nvidia supporting Windows 8, they might’ve felt they need Intel as insurance. But it all feels like a political decision.
Expecting 8 core 2.5Ghz ARM chips to consume 5x less than an atom chip is a little too optimistic, i think. As performance increases, power consumption also will increase. How much power can be saved just with process improvements?
He said cost 5x less
Lucien, I’m with you on a lot of your doubts and want to dig into the deets on what ARM will have on offer as well. There’s also the thought that anyone would be nuts to want to go back to an Intel-dominated client world. So I tend to favor the political angle as well :)
Intel has basically been sitting on its laurels with Atom (for far too long) and now they pay the price. I remember when Pinetrail was delayed back in 2009 and the same delay is happening again with Cedartrail. It would be nice to see some 22nm quad-core parts hit retail by 2013, but for that to happen IMO Intel is going to have to prioritize their low-end – something that the company has never done.
Intel also made the big mistake to place Atom on a 5-year architecture cycle. It doesn’t take an Einstein to see that mobile tech moves a lot faster than that, and even a change to a 2-year cycle may not be enough to compete with rivals such as ARM. Slate sales have been biting at the heels of netbooks this year. If MS does release Windows 8 on ARM next year, and Intel fails to impress with Cedartrail’s successor, the Atom will face some very stiff competition. And that doesn’t include what AMD plans to do with it’s Brazos platform.
Stacey, several sentences need to be edited:
Intel is very serious about low power chips, although it won’t have them “for” until 2013.
The company showed off the long-rumored Haswell chips at its developer forum on Tuesday, which it says can “can” run all day…
That question won’t be answered today, but “Intel’s is” in the very least…
However, it’s unclear where vendors such as Nvidia, Qualcomm, Samsung, Apple “(a aapl)”…
Thanks!
Does “all day” means 8 hours or 24 hours? it’s a big difference
Andy, the voice of reason.
Intel expects to enter the market in 2013 (or 2015, as internal documents and top analysts predict) with something less than A) a market disruptor, B) a disgustingly cheap, competitive equal, or C) an amazingly advanced alternative. That business model sets them up for 1-15%market share of the mobile market. AND NO MORE.