Intel discontinues (sort of) Atom CPUs (relevant to pyra and competitors)

[vcoleiro1]

Saw this story on Lilliputing, it seems from rumors, that Intel are planning to kill off their low powered SoCs in that Cherry Trail will be the end of the line. To me this makes no sense, Intel SoCs were starting to make some real headway in the cheap tablet space and getting the Intel name out there as an alternative to ARM for tablet SoCs. They invested a lot doing it , and I'm sure that they made no profit from it to do so, but now that they did it, why turn around now. It seems to me they got more than half way over the river and now decided to turn back.

http://liliputing.com/2016/04/reports-intel-is-killing-off-low-power-atom-chips.html


Ok, I now see there is another thread on this same topic in another section. Feel free to delete this thread mods

 

[Askarus]

Ok, I now see there is another thread on this same topic in another section. Feel free to delete this thread mods

Could you post a link to the other thread please ?

 

[rygD]

And can we get this thread locked, or maybe merged with the other one?

 

[Klumpen]

And can we get this thread locked, or maybe merged with the other one?

Yeah, would be a good idea otherwise more people might post in it.

Ooops!

 

[asimov-solensan]

Just read this and thought people might find interesting

http://www.pcworld.com/article/3063...-tablet-markets-after-cutting-atom-chips.html

Summary: Intel won't design more CPUs aimed at samrtphone and tablet world.

I get two points from this:
1 We wont have an x86 Pyra (not that i cared anyhow).
2 We wont see any relevant iteration for devices like the GPD win. Which will make this a unique device for a long while.

I'ts still possible to get an x86 from another company, and even get a new suitable CPU from Intel. Just seems highly improbable at this point.

 

[TeDaDeS]

Well, it states they don't see the tablet market as a different market anymore. It will just continue to use the Skylake Core M instead of the Atom.
And the Atom line still exists so you can still put one of those in the Pyra if you really want to.

 

[comradekingu]

I saw they were upset about it in the GPD indiegogo comment section. I didn't contribute, so I couldn't post a link to the pyra preorder.

Tablets has never been a market to intel, they failed, and then kept failing.
Throwing even their future atoms away means they are done with pouring money into their mobile division.
Maybe they could have something that scales down into a high-end handheld at some point. I don't see how their roadmap core M is it.

 

[Chip]

Tablets has never been a market to intel, they failed, and then kept failing.

Just the opposite - the atom chips were too successful.

Intel has wanted into the mobile space for forever. The last couple generations of atoms finally managed it, though a combination of actually being decent chips, and being dirt cheap. Today you can buy an atom-based tablet, convertible, nettop, etc from any of a hundred manufacturers. But in trying to oust ARM as the go-to chip for inexpensive tablets and 2-in-1s, intel was pretty much giving atoms away for cost (under cost, according to some estimates). The line has been massively successful in raw numbers, but isn't actually making the company much money.

What's worse, atom is eating the lunch of the core M division. The atom chips wholesale for $20-40, while the core M chips are roughly ten times as expensive ($250-350). They're significantly more powerful, sure. But they're not ten times more powerful, and atoms are good enough for basic mobile computing. Unsurprisingly, more people are willing to pay $200-300 for a good-enough atom device than are willing to pay $800-1200 for a core-M device - especially when 95% of them will spend 95% of their time with that device browsing facebook and playing farmville.

My personal theory is that this is simply a rebranding (and repricing) maneuver. I bet we'll see the exact same goldmont chips that were going to be sold as broxton atoms instead sold as as apollo lake celerons. This ditches the atom name (which has never quite lost it's early netbook stigma), and allows intel to raise prices on the sly. People would be up in arms if the next generation of atoms cost twice as much as cherry trail, but if you stick a "celeron" label on those exact same chips, then charging $40-60 for them sounds perfectly reasonable.

Mark my words - there will still be ~2W chips in the goldmont family fit for the pyra or GPD win2. They just won't be called "atom", and they won't cost <$30.

 

[Exophase]

Just the opposite - the atom chips were too successful.

Intel has wanted into the mobile space for forever. The last couple generations of atoms finally managed it, though a combination of actually being decent chips, and being dirt cheap. Today you can buy an atom-based tablet, convertible, nettop, etc from any of a hundred manufacturers. But in trying to oust ARM as the go-to chip for inexpensive tablets and 2-in-1s, intel was pretty much giving atoms away for cost (under cost, according to some estimates). The line has been massively successful in raw numbers, but isn't actually making the company much money.

You can easily buy Atom tablets and convertibles, largely thanks to Intel selling them for next to nothing. But tablets are not where it's at for mobile, they're a sliver of phone sales and tablet sales have started heavily declining. So they're not a good bet for the future.

Intel never really made it into phones and that was disastrous for their prospects. Their cellular development for instance was not really guided by tablets (yes some tablets have LTE and whatever but they're a relatively small percentage and are about as well suited with external modems)

It's kind of similar to what happened with nVidia.. who ended up having to pretty much write off their cellular acquisition. Intel's position of "we'll focus all our energy on 5G instead!" is a lot more bold but also pretty risky, and if they fail at that too I'm sure it'll be over.

My personal theory is that this is simply a rebranding (and repricing) maneuver. I bet we'll see the exact same goldmont chips that were going to be sold as broxton atoms instead sold as as apollo lake celerons. This ditches the atom name (which has never quite lost it's early netbook stigma), and allows intel to raise prices on the sly. People would be up in arms if the next generation of atoms cost twice as much as cherry trail, but if you stick a "celeron" label on those exact same chips, then charging $40-60 for them sounds perfectly reasonable.

Mark my words - there will still be ~2W chips in the goldmont family fit for the pyra or GPD win2. They just won't be called "atom", and they won't cost <$30.

Broxton had integrated modem and it looks like that entire effort - 4G on Intel 14nm - has been killed. Apollo Lake will still have Goldmont, Goldmont is the name of the CPU and not the chip. But the big question is, will they actually optimize the SoC to be able to do < 2W SKUs? If you look at this generation, we have Cherry Trail for tablets and Braswell for cheap laptops and desktops. Both have the same Airmont CPU, but the lowest TDP for Cherry Trail is 2-3x lower than that for Braswell, even with comparable clock speeds. Part of it is because of peripherals, and Apollo Lake will at least support LPDDR3/4 (but then again, so does the latest Core M). So I guess we'll see.

But unfortunately, I'm hearing from people who have a lot of connections with Intel that tablet Atoms are done.

 

[AJLond1]

So then what chips would any devices with Intel SoCs use post-Cherry Trail?

 

[fusion_power]

Mark my words - there will still be ~2W chips in the goldmont family fit for the pyra or GPD win2.

That's actualy good.

They just won't be called "atom", and they won't cost <$30.

That's actualy bad. Why would Intel do that? Rebranding is old business but usualy comes with also lowered prices to still sell out old crap under new name, but raising prices for no reason? I still really count on CHEAP x86 SoC's to put in Pyra-like Handhelds.

AMD, where are YOU actualy?

 

[comradekingu]

ARM or MIPS.
Or face having worse numbers in terms of overhead to work with, which further places it in enthusiast and quirky territory

Intel does not care about their customers. They care about their investors, and will do anything in and outside of their power to please them.

Edit: Amd has understood that it doesnt need to play second fiddle to everything intel is doing. Like their laughable low-end offerings have been.
Its a bit of a dwindling and tough business to begin with. They want servers, casual games, and in the future high-end.

AMDs ARM-chips is a feasible way to contest Intels budgets.

 

[Chip]

Why would Intel do that?

Because as I (and the article) said, intel was losing money on the atom line. They were selling the chips for as little as $17, and that wasn't enough to keep the division afloat. They need to raise prices, and figured rebranding was the easiest way to get away with it.

AMD doesn't have anything in x86 with a low enough power envelope to cram into a pocketable device. Their ARM chips are intended for low power servers, and there are obviously more than enough ARM chips from other fabs to satisfy anybody willing to go that route. I wouldn't expect anything game-changing from that camp any time soon.

 

[fusion_power]

Well, for our needs, even ARM has not much selection when it comes to small batches of available SoC's. That's why I hoped for the current Atoms. If they will be only rebranded and still be available for an fair price, this would be great. Because CoreM & Co. are not an option at all here.

 

[SNESFAN]

Truly a sad day when looking to the future. The recent Atom chips were quite good for their intended purpose. The race to the top (total power) was complimented well with the race to the bottom (power consumption and price)

hopefully its just the line they are killing and not the concept

 

[EvilDragon]

True, but the market is hard.

For Android devices, it doesn't really make sense getting an x86, unless it would have major advantages (speed or power usage) which it didn't.

As a gaming device, most people expect it to run current games - which is doesn't (see here: http://www.notebookcheck.net/Intel-HD-Graphics-Cherry-Trail-Benchmarks.140902.0.html)
It's stupid thinking that a mobile device could have the same power as a 600W desktop PC, but people expect it nevertheless. That's the curse when you offer compatible platforms...

The only market left is something like the Surface, where you can do some office work on-the-go with a clip-on keyboard.
And the surface didn't use the CherryTrail...

So yeah, I can see why that attempt from Intel failed. There's not a huge demand.

 

[Chip]

And the surface didn't use the CherryTrail...

The surface 3 uses an x7 z8700.

 

[EvilDragon]

The surface 3 uses an x7 z8700.

Ah, right, thanks. I confused those with the Surface Pro series.
I wonder how many bought the Surface... most probably got the Surface Pro.