Google May Soon Design Its Own Chips with ARM

[JMcEntegart]

Is Google about to add another product to its already vast and varied portfolio?

Google May Soon Design Its Own Chips with ARM : Read more

 

[house70]

I just hope Google is not going to become something like Apple; proprietary everything and so on. Servers CPU, OTOH, would be welcomed.

 

[JD88]

All of the other articles I've read about this don't say anything about the consumer market. This is for servers only.

 

[PhilFrisbie]

I would also tend to believe this is for servers only. Using lower powered ARM processors could save them many millions in energy per year.

 

[derekullo]

Google was founded as a search engine that gets the majority of its revenue from ads.
Designing a phone is not that much more complicated than building a pc . You choose which parts you want to go in your phone and have a manufacturer put the phone together and you call it your phone ie Google Nexus. Asus, HTC, Samsung and LG manufactured previous google nexus phones and tablets. All of which have long histories manufacturing electronics.

It's a huge jump to say i know which parts are good in a phone to I know how to make good parts. I'm not saying google doesn't have the money to research ARM cpu. Google just has 0 experience in designing cpus or electronics in general.

tl;dr?
Google may be shooting itself in the foot trying to build its own ARM cpu

 

[gm0n3y]

Given the vast size and number of Google's server farms and their highly customized nature, it would make a lot of sense for them to get more involved on the hardware side.

 

[falchard]

Microsoft thinks its a wonderful idea for Google to design its own chips and alienate all their partners.

 

[ZolaIII]

This actually have much sense! It's economically justified for Google to make servers giving in consideration they one needs. This will ensure a future for Arm servers & actuality there is no reason for them not to make consumer SoCs. It's not that hard to make a soc based on reference design with licensed parts! Tricky part is making your own costume designs bat consider that Motorola did make it's own costume SoC based on Qualcomm design already. To do this Google need to bay some actual chip maker & not a very large & overpriced one. If i whose in there shoes i should bay VIA to achieve this. There are many reasons why VIA & i will mention only some: it's well known (as a brand), it proven to be competitive with emerging Chinese chip makers, it have experience with x86 systems which is important in integrating additional infrastructure in server parts (buses, controlers, switches...), have in the house build cheap & good WiFi & Bluetooth solutions & many others. Bat Google must stick to reference designs & make licensing firms as much as possible cooperative with community with open drivers. This really can be beneficial to users in many ways & competition is always good thing. They can work with Arm to improve future designs & they can make much better analytics what is really needed & desired from final users then anyone else & this will certainly help them in optimizing software solutions (& this is major problem today).

Every ¢ that open solution steal away from closed ones is actually good thing for the future of computing & necessary one in the world of market economy for them to compete!

 

[suau]

It's a huge jump to say i know which parts are good in a phone to I know how to make good parts. I'm not saying google doesn't have the money to research ARM cpu. Google just has 0 experience in designing cpus or electronics in general.

tl;dr?
Google may be shooting itself in the foot trying to build its own ARM cpu

Motorola has experience. Google has Motorola. Google has experience.
Also the title is misleading, what the author probably wanted to say is: "Google might be working with ARM on a customized Chip design". Now it makes perfect sense with regards to the Android platform:
Android devices already have certain hardware requirements (e.g. Bluetooth, Accelerometer etc...)
Slamming it onto a single chip/chipset has one huge advantage:

Unified implementation of basic hardware requirements (which offer no distinction for individual hardware competitors)
> unified driver implementation
>> beat fragmentation
>>> simplify and speedup firmware updates for manufacturers
>>>> end users happy

now open source the drivers:
+ no more tragedy like AOSP + JBQ
+ custom ROM builders happy
+ improved drivers

With the wave of accessories coming to the market this is going to be HUGE advantage,
Just take Nike+ FuelBand SE as an example : according to Nike there is no app for android because

“Bluetooth LE (low energy) is relatively new, and on the Android operating system there are so many devices running different versions of the OS. It is not a stable technology”

sounds like a cheap excuse to me, but having the same BT on every device would make things definitely easier.