According the the article, TI is "eliminating 1,700 jobs in their mobile processor business unit" and will "stop developing new chips." You call it refocusing, I call it to terminate with extreme prejudice. Either way, it's the end of the line for OMAP CPUs.
OMAP is not just mobile processors. "...TI wants to sell its OMAP processors in markets that require less investment, like industrial clients like carmakers." So no, they are not shutting down OMAP, they are shutting down the OMAP lines that feed consumer electronics.
Their future hardware IP design work is effective(ly) dead. Embedded on OMAP has always been a derivative of the R&D invested for the mobile market. You can also look at the Sitara products which are respins of OMAP IPs with a few needs for embedded.
that makes sense, because almost no application of OMAP really used all it's features (notably embedded DSP core), only significant consumer/mobile application I can think of are Nokia's phones based on Symbian EKA2, which are effectively dead now (and OMAP combined with EKA2 was significant competitive advantage for Nokia when they wasn't, as that enabled them to replace their ARM-based custom SoC with OMAP and build smartphones essentially derived from feature phone hardware without dedicated baseband CPU). Before that (early 00's) there wasn't any other real SoC solution, as most other mobile processors required some external (mobile-specific: touch screen, power management...) support circuitry and OMAP previosly filled this niche, now it seems that there isn't any real makret for OMAP SoC's as nobody is interested in building anything that leverages OMAP architecture, because random mobile SoCs paired with some generic baseband are cheaper and consumers does not seem to care about battery life.
It does not make sense to develop hardware which nobody really uses and try to be price competitive with cheaper designs that are adequate for most customers.
The Blackberry DevAlpha A&B devices are OMAP 4460 & 4470 respectively. The battery life is remarkable. The performance on this dev alpha really puts tegra3 to shame.
The big win for me for OMAP5 was going to be using the ISP pipeline for vision and feature extraction especially with the low latencies provided by qnx. Linux is not even in the same league.
I am very disappointed about this announcement as this group consistently created industry-leading innovations. When I first heard they were trying to find an acquisition partner I also assumed Amazon would pick them up. I wonder what factors prevented this?
that makes sense, because almost no application of OMAP really used all it's features...
I'm not sure why you say that. All accelerated video encoding/decoding is done on the DSP. This is in contrast to the Freescale i.MX5 series, for example, which has dedicated hardware blocks for this.
The OMAP4 has dedicated video encode decode, dedicated "still image" processor, and a general purpose DSP (which isn't used for the most part and would be a lot slower than the dedicated hardware units).
EDIT: The thing though is that its footprint is so small that it doesn't matter much.
Indeed, The article itself confuses the issue too, mentioning that fact while also referring to TI looking to sell the entire division, as if mobile devices were the only market. Its worth noting that TI still manufactures ARM-based chips for more complex systems, which may be part of their motivation for the OMAP move.
Also, completely irrelevant to the point, TI is based in Dallas, not Austin.
> They are not shutting it down,
"refocusing" = cutting jobs ans shutting down plants ( and they do cut 1700 of them ), refocusing is just the expression suited for the "yesmen".
Samsung. Exynos 5 is powering the Chromebook and Nexus 10 after all. Allwinner is popular among the low end Chinese stuff, so there's always the chance that they'll expand into higher end SoCs.
AFAIK Freescale is taking the same path as TI, canceling i.MX7 and exiting application SoCs. ST-Ericsson has a nice A15/Rogue SoC in development, but it remains to be seen if it'll be popular among smartphones/tablets.
Do you have any hard news links discussing Freescale's roadmap? I'm about to commit a design to i.MX6 and if that's the end of the road, that's useful news.
I believe there are over a dozen ARM chip makers out there. TI seems to be out of the mobile market. I hear Freescale will be soon out, too. It probably won't be long until ST-Ericsson is out as well.
So out of the popular ones we're left with Qualcomm, Samsung, Nvidia, Broadcom, Mediatek, and maybe a few others that don't operate in western markets much.
But it's quite normal for this to happen. It happens in all mature industries. Only a few big players remain, that have out-competed everyone else out of the market. The ARM environment will still remain much more competitive than the x86 one ever was.
There's also Apple which now makes its own chips, and while it doesn't sell them to others, it's still a competitor. As competitors to the ARM chip makers we might even consider the indirect ones (other architectures) like Imagination's MIPS processors (in the future), and Intel's mobile processors (also in the future, as I don't consider them very competitive right now).
Other ARM chip makers that don't compete in the mobile market include Marvell (servers and smart TV's), Calxeda (servers) and Applied Micro (servers).
The ARM market will be alive and kicking for a very long time.
Intel's been working on mobile processors for phones. They've had a few design wins for their single-core 2GHz Atom. No phones you can buy in the USA though. Reviews for released phones like the Motorola Mobility RAZRi are decent, but didn't seem to knock anyone's socks off. If they develop good LTE integration (they bought part of Infineon a few years ago) and if their management stays committed to this, they may have a chance. Intel was previously a big backer of WiMAX.
Marvell, who bought the PXA line from Intel doesn't have chips competitive with the current generation from TI, Freescale, etc. I don't think Broadcom has anything competitive either, but they don't have much publicly available information.
You've also got Freescale. I don't know if their latest generation (the i.MX6 series) is currently shipping, or how good the OS support is. I haven't heard of any major design wins for them recently, if anyone knows, please post it. If you are focused on the automotive market, this might be a good choice because they typically have CAN bus support.
Then you've got Samsung of course, though you'll be competing against their own products. Apple doesn't sell their chips. That's about it for the majors.
You've got other mid-range chips from low-cost vendors like Allwinner, but I haven't heard of anything from those guys which will allow you to make a leading-edge phone or tablet.
Late addition: According to their website, MediaTek also has a mobile app processor, but it is ARM11 based.
I'm prototyping on an i.MX6 Quad dev board from Boundary. It's nice, I'd call the chip an incremental step from the i.MX53. Trying to benchmark it against a Marvell Armada 610 and it's no contest.
Allwinner is a comparable ARM core, but I wouldn't design it into anything I'm building for more than six months at a time. It's like all of the other noname chinese cores (Ingenic is the other one that comes to mind, but they're all MIPS). They just come and go with no warning.
