Intel Decides To Let Go Of Broxton

[Nobu]

"intel's" powervr drivers for phones were not exactly open source. maybe amd will make not only server arm.

Yeah, unfortunately. It would be nice to see AMD move into the ARM phone space, probably wont happen for a year or so, if at all. :/

you said winphone was one of reasons to want x86 phone, how should i interpret it ?

Like I said, I was referring to the OS, which was supposed to work on both ARM and x86. I wasn't aware of the existence or lack of x86 windows phones, just that the OS was supposed to work on them. Since you knew that there were no x86 (win)phones, you could have guessed I wasn't talking about the hardware, I suppose.

[starshipeleven]

Yeah, unfortunately. It would be nice to see AMD move into the ARM phone space, probably wont happen for a year or so, if at all. :/

Unlikely, the mobile ARM space is mature and many players are dropping the ball. Only Qualcomm and Mediatek still remain in there, all other players got steamrolld or simply cannot compete anymore price-wise.

Samsung does not count, the Exynos is a pet project and Samsung is fucking mindboggingly huge megacorp doing everything in asia, it's not electronics-only.

AMD should have to pull a BIG trick out of the sleeve to get into this market.

I hope that their "trick" is a decent and opensourced GPU based off their own ones. If they go Mali I'm gonna cry.

[chithanh]

Unlikely, the mobile ARM space is mature and many players are dropping the ball. Only Qualcomm and Mediatek still remain in there, all other players got steamrolld or simply cannot compete anymore price-wise.

That is not true at all. Smartphone SoCs is a cut-throat business but supports more than two companies.

Claiming that Exynos is a "pet project" is not a good reason to exclude it. Its performance can keep up very well with Qualcomm and Mediatek.
Huawei/HiSilicon is a major player and their silicon is quite competitive too.
Spreadtrum SoCs you often find in low-end phones in emerging markets.
The only company I would exclude here is Apple because they don't sell their silicon.

[starshipeleven]

That is not true at all. Smartphone SoCs is a cut-throat business but supports more than two companies.

Everyone is getting out. Marvell, Texas Instruments, SonyEricsson and others.

Claiming that Exynos is a "pet project" is not a good reason to exclude it. Its performance can keep up very well with Qualcomm and Mediatek.

I'm calling it a pet project AND excluding it as it is basically confined to SOME Samsung devices.

Huawei/HiSilicon is a major player and their silicon is quite competitive too.

Again, another giant using its own product. (much) Better than Exynos, but still not the best future imho.

Spreadtrum SoCs you often find in low-end phones in emerging markets.

Must be very low-end, I only see Mediatek in cheap crap.

The only company I would exclude here is Apple because they don't sell their silicon.

Yeah, the same reason I'm excluding Huawei and Samsung, they probably sell their SoCs, but there are none or very little buyers.

[chithanh]

Everyone is getting out.

Before it was two manufacturers, and now suddenly everyone is getting out? I still maintain that this is hyperbole and not even remotely true.

I'm calling it a pet project AND excluding it as it is basically confined to SOME Samsung devices.

That is also not true. There are devices from other manufacturers who use Exynos.

Spreadtrum SoCs you often find in low-end phones in emerging markets.

Must be very low-end, I only see Mediatek in cheap crap.

Maybe because you do not live in an emerging market? Indian domestic manufacturers (Intex, Karbonn, Micromax, etc.) have low-end Spreadtrum smartphones on offer.

Again, another giant using its own product. (much) Better than Exynos, but still not the best future imho.

Yeah, the same reason I'm excluding Huawei and Samsung, they probably sell their SoCs, but there are none or very little buyers.

Apple doesn't sell their SoC. Samsung and Huawei do, and they have not-so-small customers for them.

[starshipeleven]

Before it was two manufacturers, and now suddenly everyone is getting out?

above I said
"Unlikely, the mobile ARM space is mature and many players are dropping the ball. Only Qualcomm and Mediatek still remain in there, all other players got steamrolld or simply cannot compete anymore price-wise."

That's the same thing I repeated in the post you quoted.

And I still keep my view. Qualcomm and Mediatek are the only two real players left, the others are a minority backed by very deep pockets (Huawei and Samsung) or are locked in market that won't last long (please note that I have a Karbonn Sparkle V from Google Android One project, and it uses a Mediatek SoC, like most of the decent low-end devices around, all chinese phones from aliexpress/alibaba have Mediatek, and so on).

[chithanh]

And I still keep my view. Qualcomm and Mediatek are the only two real players left, the others are a minority backed by very deep pockets (Huawei and Samsung)

Yet the smartphone SoC vendors which you don't count have a combined marketshare of 18% according to Strategy Analytics, and most of them registered double-digit growth year-on-year: https://www.strategyanalytics.com/st...-$20.1-billion

or are locked in market that won't last long (please note that I have a Karbonn Sparkle V from Google Android One project, and it uses a Mediatek SoC, like most of the decent low-end devices around, all chinese phones from aliexpress/alibaba have Mediatek, and so on).

On the contrary. The Chinese smartphone market has stopped growing. High-end smartphones have stopped growing. The growth is in the low-end elsewhere.

[starshipeleven]

Yet the smartphone SoC vendors which you don't count have a combined marketshare of 18%

Please note, that pie chart is revenue, not market share.

It means that everyone else combined has a revenue of Mediatek alone (from SoCs sales, of course Samsung and Huawei are orders of magnitude bigger than Mediatek), or HALF the revenue of Qualcomm, while Apple has outrageously high revenue considering the tiny number of chips made (how unexpected).

and most of them registered double-digit growth year-on-year:

Doesn't really matter, top dogs have been established already. Qualcomm has won, Mediatek is second, Apple is in a league of their own as usual so they don't need to care about this (they pulled the same tick with PCs, it worked here too).

Making a good and powerful SoC requires more money in a non-linear increase, you also need expertise and market trust.

Since Qualcomm and Mediatek are already there now, by the time the low-tier guys get at the same level they will be in outer space.

On the contrary. The Chinese smartphone market has stopped growing. High-end smartphones have stopped growing. The growth is in the low-end elsewhere.

It is only temporary, eventually they will be able to afford more and more powerful smartphones, and at this point who do you think will sell the SoCs to the OEMs?

Mediatek and Qualcomm that can sell them low-end stuff they designed 2 years earlier for nearly nothing or whatever "cutting edge" coming from a smaller company?

[chithanh]

Please note, that pie chart is revenue, not market share.

Indeed it is revenue share, so actual unit share of the other vendors could even be higher than 18%.

Since Qualcomm and Mediatek are already there now, by the time the low-tier guys get at the same level they will be in outer space.

Well, comparing the 2015 numbers with 2014 ones we learn that:

Total market declined from $20.9B in 2014 to $20.1B in 2015.
Qualcomm went from 52% revenue share ($10.9B) to 42% ($8.4B)
Apple went from 18% ($3.8B) to 21% ($4.2B), which may include one-time effects as 2015 was the first full year when Apple sold large-screen iPhones.
Mediatek went from 14% ($2.9B) to 19% ($3.8B)
Everybody else went from to 16% ($3.3B) to 18% ($3.6B).

So no, the numbers don't support the idea that Qualcomm and Mediatek are pushing everybody else out of the market.

It is only temporary, eventually they will be able to afford more and more powerful smartphones, and at this point who do you think will sell the SoCs to the OEMs?

You don't need to believe me, believe respected mobile industry expert Tomi Ahonen, who recently proclaimed that "There IS NO growth at the top-end anymore" (capitalization as in original).
Android premium segment has peaked a while back, and is mostly cannibalized by mid-range devices which are "good enough" for most customers. This is exacerbated by the decline of carrier subsidies in developed markets, and the growing importance of emerging markets where such subsidies are rare.

Mediatek and Qualcomm that can sell them low-end stuff they designed 2 years earlier for nearly nothing or whatever "cutting edge" coming from a smaller company?

Google mostly prevented that by refusing devices running old Android versions the certification for GMS licenses. As Mediatek hates nothing more than porting their old SoC code to new Android versions, this means they can only sell new chips.

[Ardje]

I think a low power ARM chip with an x86 (or some subset) co-processor that could be powered down when not in use (something Intel definitely could pull off) would've been an interesting combination.

That has been around for years, as each and every server board runs linux on an arm cpu as motherboard controller and remote KVM. That cpu cannot be turned off except by unplugging the server.
So yes, all servers in the microsoft cloud at least run linux on the motherboard. As far as I know Google doesn't use server boards.