Intel has been working to reduce power consumption while improving performance across its processor lineup. The company’s upcoming 4th-generation Core processors, code-named “Haswell” are expected to offer up to three times the graphics performance of today’s chips, while using less power.

But it’s not just the expensive Core family chips that are getting a performance bump. Later this year the first low-power Intel Atom chips based on the new “Silvermont” architecture are expected to hit the streets.

They’re expected to be up to twice as fast as any Atom chips released to date, while consuming less power. We should start to see Android and Windows tablets with Silvermont-based chips in the second half of 2013. The first smartphone chips based on the new architecture will follow in early 2014.

Bay Trail

Intel has been talking about its next-gen Atom chips for months, but today the company started really digging into the details of what makes its new processors tick. And in the process, Intel has shown that it’s taking its low-power chips more seriously than ever.

The first Atom processors hit the streets in 2008, and while Intel gradually added support for multiple cores and better graphics performance, the truth is that an Intel Atom Clover Trail chip from 2012 isn’t really all that much better than a first-generation Atom chip from 2008.

By moving to a 22nm Tri-Gate manufacturing process, Intel has been able to improve performance while reducing power consumption. According to Intel, the new chips may offer as much as 3 times the peak performance of today’s chips while using the same amount of power — or the same level of performance while consuming 5 times less power.

The new chips won’t support hyperthreading, but they don’t need to, since they can actually support multiple cores — the first Bay Trail chips for tablets based on Silvermont are expected to be quad-core processors. Ultimately Intel says Silvermont will support chips with up to 8 cores.

They’ll also feature Out of Order architecture which will let them perform single-threaded tasks much more quickly. Turbo Boost has also been improved, which means, for example, that a 2 GHz Atom chip with a Silvermont core will actually be able to run at speeds even higher than 2 GHz at times as long as there’s no risk of the processor overheating.

Silvermont is also a 64-bit chip, although it’s not clear if that means every product with a Silvermont core will actually have full support for 64-bit software.

AnandTech’s Anand Lal Shimpi has posted a very detailed overview of what’s new in Silvermont, and he estimates that a tablet with a Bay Trail processor will offer performance similar to what you’d have seen from a MacBook Air with an Intel Core 2 Duo chip in 2010. That’s a huge step up from the Atom chips we’ve seen over the past few years.

Of course, Atom performance still falls far short of what you’d expect from the latest Intel Core processors. But Atom chips use significantly less power and will likely sell for much lower prices — this is the chipset that’ll probably be used for those $300 Windows 8 tablets we keep hearing about.

Silvermont-based chips will also likely power future tablets running Android.

Now that ARM-based chips like the NVIDIA Tegra 4, Samsung Exynos 5, and Qualcomm Snapdragon 600 are pushing the boundaries of what a non-x86 chip is capable of, it’s about time to see Intel stepping up the performance of its low-power chips.

While Bay Trail tablets should be available by the 2013 holiday season, we probably won’t see Silvermont-powered smartphones with “Merrifield” processors until early 2014. They’ll go head-to-head with phones featuring ARM Cortex-A15 processors, and it sounds like they could be quite competitive.

Even bigger news might be the fact that Silvermont is just the beginning. Intel now plans to launch a major update to its Atom architecture every year, just as it does with its Core processors. That means the we should see 14nm “Airmont” chips in 2014, and another major update in 2015.

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign


Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Subscribe to Liliputing via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 9,507 other subscribers

11 replies on “Intel introduces Silvermont architecture for next-gen tablet, smartphone chips”

  1. The original Atom was a product born out of contempt for us the consumer!!! It was a mediocre product from the start designed not to interfere with the higher margin Core I series…you grew frustrated with it being so poor on all fronts CPU/GPU/chipset and then you just spend again to get the more expensive product, Two sales instead of one!! Now that ARM and AMD are offering very competitive products, in the case of AMD the new TEMASH/KABINI duo, INTEL now is forced to take tho whole low cost/low power sector much more seriously, I for one did not appreciate the contempt and will steer my devices purchases towards ARM/AMD.

    But what I do will not matter as much as device makers who likely feel the same way and much prefer to live in a world where there are choices…and monopolies are held at bay by alternatives. We ought to remember this when our next device purchase comes up…we have the choice to keep the competition healthy!

    1. Yes, competition is good for consumers but the bottom line for most consumers is getting products that do what they want the best and that’s not always the company someone may prefer.

      Besides, this statement you made is wrong…

      “The original Atom was a product born out of contempt for us the consumer!!!”

      No, the ATOM was not born out of contempt for the consumer but rather because consumers were demanding low cost solutions.

      Projects like OLPC and the initial work on the ATOM started before netbooks were even first released. It was only after netbooks became popular that the limits on the ATOM were imposed and they stayed that way mainly because there was no competition at the time.

      Intel is like the proverbial sleeping giant, unless constantly engaged then they become sluggish and slow to change, but once they are engaged then they can move surprisingly fast and push out solutions that do indeed compete well with other products.

      Like before they converted the ATOM into a mobile SoC, no one ever thought they could produce a chip that could compete directly with ARM for both cost and power efficiency, at least not without either starting from scratch or taking a decade or so of R&D, but that’s exactly what they achieved and a lot faster than predicted.

      While the upcoming Bay Trail represents a complete 180 on how they managed ATOM development, putting it not only on a similar two year type tic-toc product cycle as their Core i-Series but also pushing technology they previously only really pushed for their Core i-Series.

      Bay Trail will even have a GMA based on the Ivy Bridge HD4000!

      Essentially, they’re making up for the last 5 years and catching the ATOM up to where it should have been.

      Really, most updates are considered good for just providing say about 15% improvement but Silvermont improves the CPU performance a lot more than that and the combination of increasing max clock speed to 2.1GHz, improved usage of Burst Speed, and up to quad cores for mobile and up to Octo cores for desktop/servers easily more than doubles the performance they can offer from the present ATOM.

      And the 14nm Airmont update to the 22nm Silvermont will start production before the end of 2014, just to show how rapidly they’re going to advance it from now on!

      While also still providing cost and power efficiency that’ll compete directly with ARM.

      So, like them or not, it won’t be a easy choice for consumers to ignore Intel’s offerings…

      Meanwhile, AMD isn’t ready to compete with ARM… Closest they will come this year is with the dual core version of Temash but they’ve yet to support advance power management states that allow for features like Always Connected Standby or achieve the extremely low mw idling states that Intel will be offering.

      Like the lowest rating for Temash APU is about 0.75W idle state, just counting the APU and not the rest of the chipset.

      While even Haswell can idle down to below 100mw (0.1W), never mind what the ATOM can idle down to!

      So while the 28nm AMD APUs will be more power efficient that previous AMD products… Only Intel can compete directly with ARM at this point.

      While it doesn’t look like AMD can compete on price either for mobile devices… The present AMD Hondo Z-60 APU for PC tablets has a Tray cost of about $70 compared to the present Intel ATOM Clover Trail Z2760 that has a Tray cost of just $41 and the Bay Trail update is suppose to make the ATOM even more price competitive with ARM.

      So price advantage will likely be limited to competing with Intel’s higher end processors like the very high priced Core i-Series.

  2. Well, it looks like Intel is finally bringing their latest process and an up-to-date atom core to the party – which is good. However, even though performance and efficiency matter a lot, there are bigger questions than sheer performance which will determine Intel’s success in the mobile market:

    #1. Pricing and margins. In order to compete with the ARM vendors, Intel has to prices these chips aggressively – most ARM vendors other than Qualcomm are not making huge profits.

    #2. Intel also has to bring an integrated solution which includes LTE capability on-die – Qualcomm has been riding high and Nvidia has been suffering because of their respective capabilities in regards to LTE.

    #3. Though Intel chips are a natural for Windows tablets, Win8/RT has been struggling in the tablet space versus their ARM/Android competition. Unless we see Win8/RT tablets start to dominate (unlikely, at least in the short/mid-term), handset and tablet vendors can get better prices from competing ARM vendors because the switching costs are lower (multiple vendors support the same instruction set).

    1. Price shouldn’t be a problem other than for the OEMs… ATOM SoCs are already pretty close to ARM SoC pricing and Bay Trail will bring them even closer…

      Mind that ARM SoCs are actually going up in pricing because of design options like Big.LITTLE… Combining two sets of multiple cores does have the down side of increasing costs and overall package size.

      So it’ll depend on the OEMs on whether the devices will be comparatively priced or higher… since amount of storage, screen quality, battery size, system size, what things are included or excluded, etc. is up to the OEMs and the SoC price for either ATOM or ARM makes up a small part of the total system costs.

      Though, of course there’s the effect of OS choices as Windows, even discounted, won’t be as cheap as Android, and Window devices will tend to have higher specs… like larger storage drives to compensate for the increase space a desktop OS takes up, the greater tendency to include premium parts like WACOM digitizers, etc. can all add up…

      But if they just put Android on both then they should be priced competitively…

    2. Atom chips cost pretty much the same as ARM chips to OEMs.

      Intel is only starting to make the Atom into an SoC so hopefully they’ll integrate more and more into the chip. At least the ones targetted for smaller devices.

      Intel has put a lot of work with the help of Google to make Android work well on x86 chips (specifically Intel ones).

  3. Hmm, that’s interesting news. If Atom is capable of producing 1.8ghz Core 2 speeds… Well, that’s something much more capable than today’s units. MUCH. Wonder what the graphics on this will be like

    1. The GMA will be based on a scaled down version of the Ivy Bridge’s HD4000… So about 3x the performance of the present Clover Trail GMA.

      1. In my experience Atom products’ integrated graphics woes have been in the area of Intel driver support (even on Windows). Power means nothing if you have to treat the thing as a generic VGA controller.

        1. Yes, but Intel is improving on the driver support with their own GPU… Imagination is 3rd party and they’ve always provided Intel with bad support…

          Silvermont shoultn’t have any serious issues because it’s not only using Intel’s own GPU but since it’s based on the Ivy Bridge HD4000, it benefits from the over a year of driver development.

          Mind that Intel recently released a update for the HD
          4000 GMA that provided up to a 10% performance boost and Intel does provide Linus driver support for their own GMAs… which will come in handy for Linus users and server applications, where Silvermont scales to 8 cores…

Comments are closed.