Last month TI announced it was the first to license ARM’s next-generation Eagle core. Today, ARM is announcing the official name of that core: it’s the ARM Cortex A15.

Architectural details are light, and ARM is stating that first silicon will ship in 2012 at 32/28nm. Here’s what we do know. The Cortex A15 will be a multi-core CPU, designs can have as few as a single core but most will have 2 - 4 cores depending on their target market.

The cores will all be superscalar out-of-order designs and support Long Physical Address Extensions (greater than 32-bit memory addressing). I suspect the cores will be an evolution of the Cortex A9. The Cortex A15 will support extensions to the ARMv7 instruction set to enable hardware virtualization support (among other things).

The Cortex A15 will feature private L1 caches but a shared L2 cache (similar to the A9). The L2 cache is stated to be low latency and up to 4MB in size, although smartphones will probably see smaller versions. ARM is promising FP and SIMD performance improvements, but it isn't saying anything more than that. 

ARM is listing performance as 5x a Cortex A8 but we don’t have a good estimate vs. Cortex A9. Clock targets are as follows:

1) 1 - 1.5GHz single or dual-core for smartphones and mobile devices

2) 1 - 2GHz dual or quad-core for netbooks/notebooks/nettops

3) 1.5 - 2.5GHz quad-core for home and web servers

ARM is targeting more than just smartphones with the Cortex A15. This will be the architecture that ARM takes into the low end notebook and netbook market. That’s right, with the Cortex A15 ARM is going after AMD and Intel - it wants to fend off the impending x86 assault on its territory. 

In addition to notebooks/netbooks based on Cortex A15, ARM will also be targeting the server market with its next architecture. As Xeon and Opteron grow more powerful, so does the need for simpler, lower power consumption servers. We’ve seen some companies attempt to address this market, but expect the floodgates to open in a couple of years as ARM officially supports it. The Cortex A15 will also enable virtualization support, specifically for the server market.

It’s too early to know anything about how well the Cortex A15 will do, but it’s clear that Atom (and maybe even Bobcat) are going to have to face a threat from below just as ARM is gearing up to face the threat from above. 

Given that the first Cortex A9 designs have yet to ship it’ll be a little while before we see smartphones, tablets, notebooks and servers based on Cortex A15. Today’s announcement is simply ARM’s statement of intent. ARM doesn’t plan on staying in the smartphone market forever, it has bigger things planned.

POST A COMMENT

36 Comments

View All Comments

  • Aikouka - Thursday, September 09, 2010 - link

    Maybe they were jealous of the Itanium's obvious success with code translation ;). Reply
  • Trisagion - Thursday, September 09, 2010 - link

    I'm guessing you're right. However, I think ARM is targeting the home / small server market with this new chip rather than the netbook / PC market. If quad core Cortex A15's can offer "reasonable" performance, then this idea will take off.

    Power is a huge factor for servers and the Cortex A15's power consumption will be a lot less than an Intel server CPU and a whole lot cheaper.

    Also, Windows is not that important for servers and Linux already has ARM support.
    Reply
  • FunBunny2 - Saturday, September 11, 2010 - link

    Let's see: a server world where there's no-power ARM, no-power SSD, and really relational databases. You could run an application on a hamster-in-a-cage generator. Reply
  • Klinky1984 - Thursday, September 09, 2010 - link

    While not having Windows might be an issue, I think some people might be interested in Android based Netbooks. I wonder how much performance a 2.5GHZ quadcore A15 would provide & how much power it would use. Reply
  • seanleeforever - Thursday, September 09, 2010 - link

    i for one, don't give a crap about andriod based netbooks. and i don't know many people who would.
    simply, why do i want to carry a netbook sized device that does what smart phone does? ridiculous isn't it? netbook was able to take off due to the cheap price (ultra portables normally cost 1500+, netbook can be had for as little as 200 dollars). price and size are the key here.
    a smart phone already cost 500+, but people are willing to pay it due to the portability. now, if you gonna put that in a notebook form factor, people's exception raises by a number of magnitude and running windows is one of the requirement. i fail to see the competitiveness of a ARM based netbook (smartbook, if you well), and i will take a ULV intel chip any day of the week giving the price point.
    Reply
  • yyrkoon - Saturday, September 11, 2010 - link

    Battery life *will* be better, but not sure about the video aspect. There was a new OEM on youtube several months ago showing off just such a netbook ( ARM based ) that had a really cool, but funky modular feature. e.g. as I recall, the laptop screen could be removed from the normal flip-top configuration, and snapped back in so it was basically a tablet. Pretty neat stuff. The battery life was proclaimed to be something very good, like 8-10 hours, possibly a bit higher.

    Anyways, ARM has always had very good power saving features when compared to the x86 architecture. This is why many serious embedded implementations have to give ARM a serious thought, But I do not mean just cellphones etc. Things like remote instrument readings, or even the Micron SSD that used two ARM9 cores for the drive controller ?

    However, I think TI would probably have to embrace "the dark side", and figure out a way to get Microsoft to support their hardware on more than just Windows Mobile. Or perhaps TI does not care, and only hopes to capture a minuscule portion of the market ? Or perhaps Windows Mobile would be good enough for a netbook. That, I personally am not so sure of.
    Reply
  • descendency - Thursday, September 09, 2010 - link

    How much software (OS, programs, etc) actually runs on the ARM based chips?

    I mean, a lot of people won't sniff an ARM based computer unless it can run Windows.
    Reply
  • metafor - Thursday, September 09, 2010 - link

    All phone software currently; which is appearing more and more like desktop software (with games and everything).

    I'd guess by 2012 (and realistically, 2013 when actual devices will have this), Android, WP7 and iOS would have gotten to the point functionally that they can replace 99% of netbooks without the user noticing.

    As for performance, I think people overestimate the performance of the current Atom. The biggest reason it outperforms current ARM devices is because it's paired with such a high-performance memory subsystem while typical ARM SoC's are using 32-bit, ~200MHz LP-DDR1.

    Netbook-aiming ARM devices will almost certainly be paired with faster memory which should bring it in line with whatever Atom chip is out at the time.
    Reply
  • wintermute000 - Thursday, September 09, 2010 - link

    in the consumer space, anything in android or IOS (the apple kind) for starters, meego, linux.....

    In the server space, how about the gazillion linux servers and appliances out there?

    I'd thought anandtech readers would be a bit more up to speed on things outside of MS land
    Reply
  • Rayb - Thursday, September 09, 2010 - link

    I agree, also I don't want to take anything away from android but meego in mobile computing is showing a lot of potential. Reply

Log in

Don't have an account? Sign up now