Daniel Eran Dilger
As reported by the EE Times, TI announced earlier this week that it would shift its application processor chip business from consumer products like tablets and phones to embedded industrial markets such as automotive in a bid to stabilize sales and increase profits.
The company originally developed its OMAP (Open Multimedia Applications Platform) line of ARM "system on a chip" application processors in conjunction with handset vendors and historically sold the parts directly to phone makers, including Amazon, Archos, Motorola, Nokia, LG, Palm, RIM and even Samsung, which also builds its own chips.
However, the popularity of Apple's iPod, iPhone and iPad devices, which had already sent ripples through RAM markets and depressed sales of conventional hard drives, is now changing the market for application processors, too.
Little oxygen left in the room
Unlike most other phone vendors, Apple began designing its own custom ARM chips after the release of iPhone, incrementally acquiring a series of chip design firms (including PA Semi, Intrinsity, Anobit and AuthenTec) that has helped it build a series of increasingly customized SoCs named A4, A5, A5X and the new A6.
Apart from Samsung, which both designs and builds its own Exynos line of ARM SoCs, most phone and device vendors have bought off the shelf components, either TI's OMAP chips, Qualcomm's Snapdragon or Nvidia's Tegra. Intel is also marketing its own x86-compatible Atom chips.
Compared to Apple's blockbuster iOS device volumes, TI's OMAP chip customers have suffered a series of notable sales disappointments, including the Amazon Kindle Fire, Palm Pre, RIM BlackBerry Playbook, Motorola's Xyboard tablet and MOTOACTIVE music player, Nokia's N9, Google's Nexus Q and Galaxy Nexus.
Given its plans to exit the consumer-oriented smartphone and tablet chip market, it appears TI's next generation OMAP5, built around the ARM Cortex-A15 core design, may never make it to market. OMAP5 was in the running with Samsun's Exynos 5 to become one of the first chips to use the Cortex-A15 design.
Apple's release of the new A6 (pictured above), which incorporates some technologies of the Cortex-A15 core without using as much power, may likely have had a impact upon TI's future prospects, given the difficulty of finding high volume sales for the new fifth generation OMAP design.
Apple's big investments in silicon design
The task of keeping up with Qualcomm's range of "off the shelf" chips and competing head to head with Nvidia for a market that is increasingly being dominated by Apple and Samsung (who both develop their own application processors) is increasingly pinching the opportunities for companies like TI that develop new parts they hope third parties can sell in their products.
Apple's essentially guaranteed high sales volumes have allowed the company to invest millions of dollars to create custom, mass production designs with proprietary features (including support for sophisticated noise cancelation in the A5 that powered Siri last year) while getting rid of unused silicon that would otherwise have a negative impact on battery life. The company also has multiple tiers of products that allow it to use volumes of slower parts and get long lifespans out of new generations of chip designs.
Apple has been actively courting experienced employees at TI to help design and support its internal A-series chips. Unlike TI, which had to account for significant overhead with its chip clients related to managing documentation and support for various OMAP chips and features, Apple is the only consumer of its internally designed A-series chips. That frees the company to focus on building only what it wants and needs, and it doesn't have to communicate with third parties about the intricacies of how its silicon works.
Apple's growing staff of A-series chip developers work in top secrecy near the company's Infinite Loop campus, and recently found out that they will not be among the groups moving to the company's new "spaceship" Apple Campus 2 now in the construction planning phase.
William Gallagher
Christine McKee
AppleInsider Staff
Chip Loder
Malcolm Owen
13 Comments
This is a huge problem for ARM! This will just jack up the entry point for high performance ARM designs. Admittedly OMAP was never an industry leading design, but if we get to the point that the only ARM hadware available is custom hardware tailored and controlled by a minority of users the market in general for ARM processors will suffer. The market really needs a generally available ARM processor of this type.
Apple's release of the new A6 (pictured above), which incorporates some technologies of the Cortex-A15 core without using as much power, may likely have had a impact upon TI's future prospects, given the difficulty of finding high volume sales for the new fifth generation OMAP design.
In this case, no. Apple had nothing to do with TI's plans and OMAP5. TI cannot find and OEM or carrier to commit because unlike everyone else, their SOC does not incorporate an LTE processor, which everyone is now basically requiring. This is why OMAP has ended up in tablets of lates. As a result they don't really have a plan forward in time to still be viable.
It is a sad day for ARM world in general though. TI provides some of the best documentation in business.
This is a huge problem for ARM! This will just jack up the entry point for high performance ARM designs. Admittedly OMAP was never an industry leading design, but if we get to the point that the only ARM hadware available is custom hardware tailored and controlled by a minority of users the market in general for ARM processors will suffer. The market really needs a generally available ARM processor of this type.
Well, yeah, but as the article points out, only Apple and Samsung are even doing custom SoC's and only Apple is doing true custom CPU's. The majority of ARM chips available are standard ARM designs. Besides, now Apple has gone custom, whether it's A-8, A-9 or A-15 doesn't really matter anymore. It's the instruction set that matters.
Well, yeah, but as the article points out, only Apple and Samsung are even doing custom SoC's and only Apple is doing true custom CPU's. The majority of ARM chips available are standard ARM designs. Besides, now Apple has gone custom, whether it's A-8, A-9 or A-15 doesn't really matter anymore. It's the instruction set that matters.
That's incorrect. Samsung's Exynos processors are Cortex A licensed, not custom. Qualcomm's Snapdragon is the custom chip along with Apple.
In this case, no. Apple had nothing to do with TI's plans and OMAP5. TI cannot find and OEM or carrier to commit because unlike everyone else, their SOC does not incorporate an LTE processor, which everyone is now basically requiring. This is why OMAP has ended up in tablets of lates. As a result they don't really have a plan forward in time to still be viable.
It is a sad day for ARM world in general though. TI provides some of the best documentation in business.
TI's support was woeful. There's been some really serious smartphone bugs over the years that can be traced back to TI not bothering to fix important defects. Unfortunately, the alternatives used to be worse - get stiffed by Qualcomm or going for a semico whose support is even worse.
These days most manufacturers are going down the "get stiffed" route.
It's a shame to see TI go through. OMAP2 and OMAP3 series chips were the best around.
Well, yeah, but as the article points out, only Apple and Samsung are even doing custom SoC's and only Apple is doing true custom CPU's. The majority of ARM chips available are standard ARM designs. Besides, now Apple has gone custom, whether it's A-8, A-9 or A-15 doesn't really matter anymore. It's the instruction set that matters.
I'm not sure if it's still true but Nokia used to do a lot of their own chips at the low and mid-ranges. It's one of the main reasons why they could undercut their rivals by so much.