We recently had an opportunity to chat with Intel about its Atom Z2460 processor, the company’s first SoC hardware for modern smartphones. Intel has been criticized for being late to the party, but Sumeet Syal, Director of Product Marketing for the ultramobile Atom processors, believes that Intel’s timing is about right.
When released later this year in initial phones, the Atom Z2460, formerly code-named Medfield, will debut with a single-core, hyper-threaded 1.6 GHz CPU that is built in 32 nm and is combined with an Infineon modem as well as Imagination Technologies graphics. Intel has already made claims that it will lead the performance space with this chip and especially the graphics engine, which clocks at 400 MHz, about a third higher than the targeted competition at this time.
Our Q & A with Sumeet Syal, Director of Product Marketing at Intel:
GBR: Medfield stands out in today’s smartphone SoC landscape because it is a single-core offering in a dual-core and quad-core world. Hardware makers are shaping a perception that more cores equal more performance. A disadvantage for Intel?
Sumeet Syal: You will see a lot of marketing around dual-core and people will talk about dual-core. However, we have a lot of research and testing on what people value in their phones. People value performance such as faster page loads. They do not necessarily value the number of cores. We also found that performance has a strong association to Intel as a trusted CPU maker that knows how to achieve performance. The dual-core marketing hype is not going to stop, but the consumer who cares to understand the performance element will see the value in our hyperthreaded chip. That said, we are working on dual-core SoCs and we know how to build those cores for the optimized smartphone environment.
GBR: Intel first began talking about a smartphone Atom chip when the initial 45 nm/130 nm Atom/chipset combination was released in 2008. Moorestown was supposed to be this product and was slated for a 2010 release, with some hopes to win over Apple as a potential customer. Medfield arrives more than four years after the initial Atom – an eternity in chip design. Why did it take Intel so long to come up with a competitive SoC?
Sumeet Syal: The challenge was integration. Moorestown was a 65 nm and a 45 nm part. There was a Northbridge and a Southbridge. We were optimizing for the longest time on power and it all came together on 32 nm based on high-K metal gate technology that delivered very low leakage power. I would say that Moorestown got the architecture pretty robust, and we took all that learning, put it on 32 nm and got the best of both worlds. That took us a while, if you will, but I would not say that we missed the market. The product is very robust.
GBR: When designing a smartphone processor today, what are the requirements for a successful product?
Sumeet Syal: If you ask my manager, Mike Bell [VP of Intel Ultra Mobility and joined the company from Palm and Apple], he would say that Intel knows how to build silicon, but a platform comes together when you build software that takes advantage of the silicon. Mike has brought in a bunch of expertise from the outside from several companies with the purpose of optimizing software for the hardware architecture. For graphics, for example, he expected the hardware to just deliver more performance. You see that in the smartphone reference design we have shown at CES. That reference design, by the way, was not just a reference design. It is finished product.
GBR: Can you give us any idea how many Medfields have shipped so far?
Sumeet Syal: There are several OEMs that are testing. We have built thousands of this part.
GBR: You previously said that Medfield would launch in commercial products in H2 of this year. How would you define a successful launch?
Sumeet Syal: We are not putting numbers out there and we did not give a specific timeline other than 2012. Motorola did not provide a launch date, but Lenovo said it would be shipping a device in H2 2012 in China. We cannot disclose whether there will be U.S. products for this year.
GBR: There has been some confusion about Android software support in Medfield. Can and will your chip run all Android applications?
Sumeet Syal: Medfield x86 has optimizations for Android Ice Cream Sandwich specifically. We have a binary translator that takes everything in the Android space for ARM and runs all apps. So, all Android apps that people are used to running, run just fine on x86. Consumers will not see a difference in app compatibility to arm and claims that Medfield will not run Android applications are just not true.
GBR: Why did you decide to go with Android? You had a bet on MeeGo before and there were other choices such as WebOS that would have enabled Intel to potentially build its own ecosystem.
Sumeet Syal: It was the sheer market momentum. The momentum had shifted to Android: We needed a partner fast and joined hands with Google at IDF. If you look at the market, there are two dominant mobile operating systems – one is closed and one is a little more open. Android made a lot of sense to us.
GBR: There is no interest in WebOS?
Sumeet Syal: I will say this: I came from Palm and we still decided to go with Google. As far as WebOS is concerned, HP has to tell us what the plans with WebOS are.
GBR: What are Intel’s immediate plans after Medfield to evolve the product?
Sumeet Syal: We believe that optimization of the platform is critical to provide a better competitive offering. From a hardware hardware perspective, we will be pushing the power envelope phones. In 2013, we will be offering a 22 nm product next year and 14 nm in 2014.
GBR: Thank you for the interview.