Slashdot Mirror
Intel Cuts Atom Chips, Basically Giving Up On Smartphone and Tablet Market (pcworld.com)
Intel, the marquee PC chipmaker, has long struggled to get a foothold in the smartphone market. The company, which was late in joining the mobile platform, is still playing catchup with Qualcomm and MediaTek. And it appears it's finally giving up on this ambition. The company is "immediately canceling" Atom chips, code-named Sofia and Broxton, for mobile devices, reports PCWorld, citing a company's spokesperson. The publication reports:Intel's mobile chip roadmap now has a giant hole after the cancellation of the chips. Intel's existing smartphone and tablet-only chips are aging and due for upgrades, and no major replacements are in sight. Sofia is already shipping, and Broxton was due to ship this year but had been delayed. Intel is also discontinuing its Atom X5 line of tablet chips code-named Cherry Trail, which is being replaced by Pentium and Celeron chips code-named Apollo Lake, aimed more at hybrids than pure tablets. Many PC makers are already choosing Intel's Skylake Core M processors over Cherry Trail for hybrids and PC-like tablets.The announcement comes days after its CEO outlined the company's future vision, and a week after the chipmaker let go 12,000 people.
Ha. (Sorry about people who lose their jobs though.)
1974-2016.
My first impression is that this does not look like a good day for Microsoft. Is this back to Windows RT? That worked so well last time.
Intel not Inside
Time to stock up on prepper basics: food, water, ammo, and silicon.
They were so close at catching up too. I wonder if they will continue to improve Core M's power consumption and eventually make an SOC version of it to compete in mobile market
Not surprised really, Atom seems a bit redundant between Core M and Core i3. Interesting to see which one pfSense goes with.
My prediction of Intel becoming IBM is coming true much faster than I expected.
Up and atom!
Up and at them!
Oligopolies and monopolies typically lose their competitive edge once they reach the top of the hill.
They get fat, happy, and entrenched in their ways and cannot undo the bad habits until things are so bad that they have no other choice but to change. By that time the company is usually too deep to be saved.
It's kind of like somebody not changing their bad diet until they have a debilitating stroke. However, the equivalent of a stroke in Big Co-ville is bankruptcy.
Rinse, die, repeat.
IBM was very lucky to have a decent second life in the late 90's. But it was largely because their consulting side was a mostly a new team, who were still fresh and hungry. Whether they'll get a 3rd chance will be interesting to see...
Table-ized A.I.
I got an Atom 330 back in 2009 and actually used it for about 5 years (until in 2014 I replaced it with an Athlon II from 2010). It was "good enough" once you got used to it, but obviously very low-end.
What became impressive later, was in 2013 when Haswell processors started getting into the same wattage ballpark, but needless to say, with way awesomer performance. And that's a couple generations ago: I bet Intel's own Skylake (not just ARM) is what really put the squeeze on Atom.
Maybe Intel is really getting out of mobiles, but I think it's more believable that they're going to use their much better cores in that market.
Intel doesn't view tablets as a standalone market any longer, with form factors quickly merging.
Maybe Intel is right. The market for phones is saturated (Apple knows this) and low profit margin. Tablets are a passing fad. The future is somewhere else.
I use AMD in my desktops as they're much cheaper for not much loss in performance.
I do have a Bay Trail tablet, and this news saddens me. It is the death of mobile x86. Sure x86 tablets did not have the battery life as ARM but it had one major advantage. The ability to use real desktop programs. In a closed source environment such as Windows, this was extremely important. Especially when your work involves software which requires desktop computing and does not come with source code. Likewise you never had to deal with crappy mobile ports.
(And before someone quips about touch screen, touchpointermouse turns the entire tablet screen into a giant track pad. )
Hold the phone, they are killing their attempt to unseat ARM in phones and tablets (sucking up too much power and not offering enough performance), but have a vision for chips in IoT devices, which have a tighter power envelope? Color me confused.
for some things, at least.
for audio, it was great. an x86 platform with a proper network interface, proper sata interface, expandable memory to reasonable mounts (for its application) and often there was onboard atx psu parts so you gave it a laptop 18v dc brick and that was your whole psu. totally silent, with an ssd.
otoh, the recent (last year or 2) of i3 has been so cool running, you can just use the fanless i3 variant on an itx board and have more fun.
still, the atom on the board was low cost, often fanless (more than the i3 was) and good enough for some video and any audio you could throw at it. it could be a nas server, as well.
--
"It is now safe to switch off your computer."
There was also a dot-com boom and this penguin guy to help them along in the 90's. I hope IBM had a "plan B" and I hope they adopt some stable and sane HR policies... I think that would go a long ways towards a "3rd chance"
C|N>K
Is it that difficult to make a low-power 80x86 ISA chip to compete with ARM manufacturers? I know the legacy instruction decoding is always going to take space, but I thought at this point the transistor count compared to the rest of the chip was small. I figured Intel with their leading edge fabs would be able to pull it off.
Having seen the bloom of CPU architectures in the 80s and 90s, and the effective monoculture we have now on the performance end of things, it would be interesting to see competing models and new attempts start popping up here and there. Everyone points to Apple's A-series as possibly moving to the desktop, but why just them? The world worked just fine when there was competition â" x86 vs 68k, PPC vs Alpha vs x86, etc. etc. Good things could grow if Intel would no longer be the 500-lb beast.
brwski
"Because without beer, things do not seem to go as well''
I wonder what this means for Windows 10 Mobile.
The tiny but dedicated fanbase kept insisting that an x86-based "Surface Phone" was on the way "any time now", and was going to be the platform's saviour. But now there's no Atom, and I doubt a Core M is going to fit in there anytime soon.
Does it make you happy you're so strange?
Intel was caught napping by the mobile revolution, and they were late to the party. Thanks to iPhone and Android devices, ARM is the standard for mobile.
Now, that by itself doesn't force out Intel. But ARM is very inexpensive, and available from multiple vendors. Intel's business model is to make chips that you need, that you can only get from Intel, and then charge a very profitable margin on those chips. Intel does not want to compete on price in a commodity market; that's not what they do.
So now Intel was trying to carve out a share of the mobile chip market, and it was competing against a chip design that is available from roughly six different companies. Their desired end game would be for the mobile companies to buy Intel chips, get locked in so they depended on Intel chips, and pay a profitable margin to Intel for those chips. But none of the mobile manufacturers wanted that... why would they? Why not just keep using ARM, which is getting more and more powerful anyway?
Intel basically had to pay companies to use the Atom. A few took Intel up on it, but those devices did not shake up the market at all. Basically a mobile device with an Atom was about as good as a mobile device with an ARM chip.
The only way this could possibly have worked would have been for Atom to be better than ARM, and not just a little better; it had to be so much better that it was a clear slam-dunk win, such an amazing chip that it would be worth the risk of entering into an entangling agreement with Intel (and being on the hook for Intel raising the prices on the chips). I see no evidence that Atom was really better at all than the ARM chips, let alone that much better.
So Intel is now going to stop paying companies to build with Atom, and is giving up on that whole market.
P.S. I would love a small form-factor PC running a 64-bit ARM chip with completely passive cooling and running Linux. I'd buy that. I might even buy it if it was called a "ChromeBox" and came with Chrome OS pre-installed, but it would be an easier sell if I could get drivers for plain Linux for all the hardware.
x86 looks pretty safe on the desktop for now, but give it a few years and we'll see if that's still true.
lf(1): it's like ls(1) but sorts filenames by extension, tersely
After owning a Surface Pro 3 for a few months, and subsequently selling it, I often wonder how people can stand using Windows as a tablet.
I mean, Windows doesn't have very good tablet integration still, even with Windows 10... ever try to close those tiny close buttons for your non-metro programs?
Secondly, battery life is complete shit.. 3 hours or less I got out of my Surface.
Thirdly, you still have to run anti-virus, and all that other crap just like a desktop or laptop computer.
And lastly, you pretty much still need a keyboard with Windows.. which makes you think, why didn't I just buy a laptop instead?
If you want or need a tablet for the portability, you're still better off with an iPad or Android tablet.
x86 requires more power than ARM which requires bigger batteries and produces more heat.
It's a fundamentally different architecture. ARM will never be able to compete with x86 in terms of computing power and x86 can't compete with ARM in terms of efficiency and low power.
You can never do more for the same cost of doing less.
Intel is making a good decision to just focus on what they're good at. If they want to compete in the mobile market they need to try to come up with a better ARM chip.
Work Safe Porn
What's the impact on Intel's Edison?
AFAIK the current Edison board uses Atom Silvermont cores.
Does this signal a retreat from the embedded market?
I could understand throwing in the towel on mobile, but IoT too?!
In dinosaur terms, that's like taking a pass on feathers and viviparous birth with one bad dating choice.
[Captcha: "botchers". How very apropos.]
Atom didn't have the power of a desktop CPU and didn't have low enough power consumption to unseat ARM. Atom also had an awkward position in SOHO NAS devices. My Atom based Synology can't transcode 1080p video. I'm sure Synology picked Atom for price/power consumption reasons - but in a box with 5 spinning discs, does power consumption really matter all that much?
I'm hoping that with the death of Atom, SOHO NAS vendors will migrate to more capable CPUs.
You can even have multi-cpu 64bit via nano chips in your low power box. Excellent crypto acceleration too. Oh, and AMD has a couple offerings in that space too.
intel stepping out means more room for via and amd. So... not a bad thing really.
let go. Welcome to Oceania!
Considering that the average iPads sold today is faster than the average laptop (i.e. Intel Core), the answer would appear to be yes.
http://wccftech.com/apple-a9xi...
1) Sell company to Foxconn
2) Rename Foxconn to IBM
3) ???
4) Profit!
It's not fair to characterize Atom as an "expensive failure". Intel helped make inexpensive netbooks and tablets widely available. If they hadn't done that, someone else would have and then Intel could be in a much worse position. They also seemed to have learned a lot, e.g. Skylake scales down to very low power and Atom-like cores are used in the Xeon Phi.
As usual, people see the name of a specific flavor of an SoC and think that the entire line is being spoken about.
An SoC has a number of process cores of a given types and then a bunch of other units that are specific to the application. A different application may call for a different set of units which will have a different product name.
The older example of this when people mention the sale of StrongARM as if that were the entirety of ARM development, when in reality is was just a specific ARM-based part that today no one talks about. There are plenty of other ARM products being sold.
Today we have word-salad about some x86 SOC parts being discontinued/not discontinued. Cherry Trail is being discontinued... because it is old and crappy and its replacement is now ready. The replacement is "Pentium and Celeron chips code-named Apollo Lake"... except the "Pentium and Celeron" cores are actually Atom-based. Pentium and Celeron are currently just brand names that carry no technical detail about the core they are applied to.
And with one swift blow cheap low power x86 tablets died.
Shame, I actually quite like my Dell Venue 8 Pro. Was looking to upgrade in a year or so, but thats not looking like the case anymore.
And that Chromebox should have a Matrox G450 quad video output and cardbus and integrated hotswap UPS. In otherwords a labotomized supercomputer turned into a portable desktop that moves better than a laptop.
Apps are for COWS!!!!! :-)
Bay Trail belong to the desktop line of Atom chips, that's right Atom tablets don't use a specific "tablet" version of Atom.
The reporting is weak, as is now usual.
What's really killed of is Atom for smartphones. Say good bye to Atom x3/x5 too at least. In other terms, everything that targeted Android is cancelled, what you won't be able to have is an Android x86 tablet, for which there is no market anyway. New 14nm desktop/laptop Atom will be launched in H2 2016 and tablets will simply have to use that.
In fact, with Android x86 hardware you might have found yourself stuck, only able to run Android variants. I'd say nvidia Tegra makes more sense, as it is rather PC-like.. even though it's using ARM. It wouldn't fit your use case, what it may do instead is to run real linux.
Tegra is not viable in tablets either, but it does not matter : it's now hardware for car dash, plane cockpit etc. that happens to be thrown together in a tablet as a showcase piece and niche product. x86 tablets may live on as low end netbook/laptop hardware in a tablet, and most ARM tablets are kind of smartphone hardware in a tablet.
Intel is in trouble over the next few years unless they have something backing in a back room that is truly ground breaking. I have a feeling we are about to have a very deja vu moment in the next year or so for anyone that remembers Transmeta anyway.
x86 has always been its own worst enemy.
Apple has patented a handful of things recently that will make it very easy for them to ditch Intel on their macbooks and even macs. In fact I believe they may have found a way to really make Intel and to be more specific x86 look dated.
US patent 9,311,152 awarded just this month.... "Methods and apparatuses for load balancing between multiple processing units " If you go through and read the patent its talking about a software layer that talks to a kernal on behalf of all the Systems computing units and delegates instructions to each. Its a method very much like Transmetas old code morphing software, that allowed them to run x86 code on a VLIW chip. Only Apples patent goes a step further and talks about breaking up standard CPU code instructions and feeding them to the GPU, or dedicated codec cores and digital processors on the fly. It would mean code compiled for x86 wouldn't have to change, and the hardware could use much lower power dedicated cores to tackle the math a lot more efficiently.
That patent and a handful of others they have been awarded recently like 7,015,919 "Virtualization of graphics resources" (which could allow the same software to address one virtual GPU space and use SOC + Dedicated GPUs and the system in a much better way then any SLI or Crossfire setup.
Is apple is getting ready to dump intel... if it isn't the A10 processor this year it will be the A11 the year after. It sounds to me like they are getting ready to drop a never version of OSX (likely a rebrand iOS+OSx merger) with a Kernal Translation unit (like the Code Morph kernal of transmeta days) that will access there processor that is likely going to have multiple cores that are not identical... more likely a dual or triple core ARM + 2 or 3 other single purpose cores, dedicated to things like video decoding ect. If there kernal translator works like the patent describes, it will be able to translate any x86 calls the software makes as well as gpu calls and feed things to which ever part of the core(or dedicated GPU or any other system Processing unit) they wish. If it works as the patent describes its very likely that for 95% of things people do... the A10 SOC could end up being as fast and perhaps even faster then the current high end Intel chips at a fraction of the power usage. It also means there would be nothing stopping them from dropping Full Mac computers with 2 or 3 A10 Cores... cause the kernal translator would just split up the work accordingly. A mac running 2 or 3 A10 processors with a dedicated GPU could end up being one of the fastest desktops on the market. (and it would run any osx software currently out there no need to recompile anything)
The only thing that might buy intel time to refocus at that point... would be the fact that that solution would be tied to Apple computers... and it would take some time (if ever) for MS to feel they need to do the same thing for Win10 hardware.
Marquee means a sign, while marque means a brand.
well it seems that Intel knows that a phone and tablet will never run well on x86... Someone needs to let Microsoft in on this secret! LOL
From what I recall of the Atom it wasn't good for anything. It wasn't low power enough for embedded applications, and not powerful enough for other applications. All, they were used for was cheap netbook type things.