Intel CEO Talks About Competing with Arm

Yeah, that's the same strategy you've had with Atom all along, which has succeeded in:

*creating the worst PC experience you can buy without digging an old PC out of the trash
* cannibalizing sales of real notebooks
* failing to deliver a TDP sufficient for cellphones without clocking them down to 300mhz

Nowhere within there have they created a chip that's faster AND competitive on performance per watt. They have faster chips that consume way too much power, and they have that are semi-competitive on TDP but are not as fast as ARM at the same TDP.

Scrap Atom architecture. It's NEVER been efficient, it's NEVER been fast, it's NEVER beaten anything in it's class. It's been outclassed by a low-powered core2duo since it's introduction.

Intel needs to take Oak Trail to 22nm with 4 CPUs and a 24EU HD4000 GPU...

Hmm... Well, my tiny Netbook that lasts 8 hours on a 6 cell battery and fits into my day planner would have to disagree Sting....

Find me any C2D that can do all of that.

Really intel, you'll provide the highest power at the lowest cost? Puhlease...

Thanks to your competition we can actually afford some of your products these days, otherwise we'd be payink 1k for an atom I'm sure.

I'm sorry Intel. This is one battle you're gonna lose. ARM is the most popular platform in the world for a very good reason and that's because ever since they were conceived in the early 80s as a competitor to the x86 platform they have beaten x86 based chip on cost due to minimal silicon real-estate and performance per watt. The platform is so popular that it's practically impossible to imagine a household without several ARM chips. Be it in the micro-oven, washer-dryer, or in mobile phones.

They said they would "Attack from three fronts", but I only counted 2 - (1)Atom chips in the tablet market, and (2)using their "manufacturing prowess" to out perform ARM. What's the third?

Bribery naturally.

If I'm not mistaken, TSMC fabs some of the Atom chips, so what state of the art silicon transistor technology are they using that anyone else can't also use? Heck, AMD fabs a lot of their stuff at TSMC too. Does Intel own the process by which these chips are fabbed? I mean, if TSMC has the equipment necessary to fab these chips, can't anyone "rent-out" that equipment? I guess I'm confused.

So is Intel making an ARM system??? or is this CEO from China? I am just kiddin ofcource, but it seems like ARM is just a fancy name for the new types of processors. It can still be called Pentium, but with smaller design process, and what's so great about the new instruction set anyways...

Zacate!

Intel + ARM =

In other words..

Intel + ARM = unhappy people with cost/performance ration and power usage.
Unless they somehow take Atom to 10nm intel loses.

Rate me down all you want but i'll always stay +1 i hope.

What the hell are you talking about, FloKid?

What you possibly mean is that the ARM microarchitecture will be ARM, but whatever Intel odes withit...they will call it a "pentium"?

Last time I checked a low power Core 2 uses about 20w+ when idle. Atom based on Oak Trail is supposed to use somewhere around 8W max. Quite a difference.

Performance wise, Core 2 will beat Atom. Atom isn't meant for mainstream or high performance. Its mean for UMIDs. Its the reason why Netbooks even exist and it will only get better as Intel goes lower on their process technology.




http://www.eetimes.com/electronics [...] ip-on-hold

Not quite. It didn't exactly go through and currently Oak Trail is using Intels 45nm High-K/Metal Gate technology that no one else has been able to get and as well wont have till 32nm with AMDs Bulldozer.

I would imagine that's the state of the art silicon transistor technology that he is talking about since Intel is normally the top of the transistor tech industry. Plus they can always push Atom down to 32nm which will cut power usage down quite a bit. Or skip to 22nm. that would probably devastate the competition though since they can push near 3.8GHz at about 1v.

I don't think Intel will try to crush ARM but more challenge them.




No Intel is still using x86. Just that Atom is their offering in the UMID world and if played right, can challenge ARM which will be betetr for us in the end.

As for new instruction sets, they always make stuff better. Normally they allow the CPU to perform that task better and faster so long as the software takes advantage of the instruction sets.

The article is about Intel x86 competing with ARM, let's do the math:

Lowest forthcoming TDP of an Atom SOC is like 4w, but the thing only runs at like 300mhz

Typical ARM TDP: 1w, and much faster than the above mentioned Atom processor.

Attempting to muscle an inferior design into the market with advanced fab tech is a huge fail re: comparing 65nm Pentium IV to a superior 90nm Athlon 64. Thank goodness Intel has monopoly money and undue influence on the industry, if AMD had a Larrabee followed by an Atom, they would've already already went under by now.

10nm will not be reached with current semiconductor technology. The practical limit has been placed somewhere around 20 nm.

Intel seeks to dominate another market. ARM is there and dominating because their offering is superior to the crappy Atom chips.

it was to be expected, I didn't say anything, secretly hoping Intel would try to keep the (dying) x86 architecture alive, but I knew in advance, when MS made Windows ARM compatible, that's the end for x86; and the beginning of ultra affordable netbooks!

x86 is over engineered. ARM saw that 20 years ago. ARMs run well from 1W to 10W. ARM can scale up but INTC can't scale down as well with x86 and the legacy they have to carry. ARM can just exploit multi-core interconnect to scale-up. Intel can do the same but with less overall benefit. I don't count intel out but I wouldn't buy their stock either. Intel's response seems misguided.

Intel should:
Try to open up more to competitor by licensing the x86 architecture. Why is AMD the only one making x86 CPUs ? Doing this may give turn the tables very quickly on ARM. Instead, if Intel shuts down, its potential partners will keep developing a competing technology.

Intel needs to trim down the x86 instruction set, and needs to get rid of many technologies that were created for serial, non-multimedia programs. making the x86 arch better resemble a SMP (GPU) should reduce power consumption. In addition to that, they would need to include a new instruction to differentiate x86_mobile from x86_desktop. Any app not switching the new flag on would run on virtualization mode, with the missing instructions emulated. In short, Intel has to make the legacy foulks angry if it wants to move as fast as the nimbler competitors.

As performance demands and CPU complexity increase, the x86 architecture will be more favored and ARM less. For Intel to have a presence in this sector, so it doesn't become an exclusive standard for ARM, makes a whole lot of sense. There's also the point that computing tiers tend to become conquered from below. That's how Intel, x86 and MS made it originally. And it's also, for instance, the reason why MS must compete with Sony PS. And it's also why Intel must compete with ARM.

There's a lot of confused and confusing comments on processors and processor technologies here. It seems pointless to adress them all. I just want to make a few comments. Atom is not crappy and never was. The chipset Intel forced most netbooks makers to use was OTOH very crappy and is to account for any perceived low performance/power. In terms of architecture, x86 is a very sound place to be. ARM is only able to compete in some low performance areas where very little, and very specialized, is asked from it. ARM is only a good architecture for small processors.

...And is Intel evil? Of course they are! I never buy Intel if I can avoid it. Eventually I suppose I will have to, when they have accomplished their monopoly, but until then they don't get any money from me. So I'll buy ARM of course. I just wanted to explain that Intel's position to compete is actually pretty good.

Nope, it's neither silicon real-estate nor performance per Watt which have made ARM so popular. One reason is that it got a real foothold in the market, which was improved during the time Intel owned it. This made it a start of a standard. During this time most competition also elected to quit, so it's not like there were many choices around either. Why it's popular today is because it's open. Anybody can roll their own, in their own SOC, like Apple did, and now nVidia also do.

Sorry, I meant it's licenced to anyone.

The problem is with the x86 ISA.There is no way to compete with an RISC ISA like ARM in power consumption with the inefficient x86

Intel could license ARM as well and manufature the chips with their latest technology. If other companies could make dollars on making ARM on license, sodo could Intel. Would be more profitable than putting costly R&D money on keeping an old architecture alive.

@kronos_cornelius - to your point on mobile_x86 versus desktop_x86. Resorting to desktop_x86 emulation on a mobile platform is only a stop-gap measure (but an absolutely necessary stop-gap measure). The goal is to move toward FATTER applications on the server-side (app-store) and downloading on the right target binary.

Whoever in this thread think that x86 is just fine for mobile, they have not understood the efforts (and Si area) that Intel devotes to decoding and supporting the long legacy of x86. An objective analysis between ARM and x86 ISA would show that much more energy (area) is devoted to decoding an x86 than an ARM. That's a non-started in the mobile space for a long time (until we get ultra-capacity batteries). Even when mobile devices are less power limited, there will be a strong motivation to keep the power-compute efficiency envelop very small.

[citation][nom]spqrusa Whoever in this thread think that x86 is just fine for mobile, they have not understood the efforts (and Si area) that Intel devotes to decoding and supporting the long legacy of x86. An objective analysis between ARM and x86 ISA would show that much more energy (area) is devoted to decoding an x86 than an ARM. That's a non-started in the mobile space for a long time (until we get ultra-capacity batteries). Even when mobile devices are less power limited, there will be a strong motivation to keep the power-compute efficiency envelop very small.[/citation]
In fact very little area is devoted to support x86 legacy. So little, it's completely irrelevant. Area is devoted to decoding though. That is correct. But that is not to disadvantage for the x86 architecture,.. - When we consider higher levels of performance. In term of raw performance, Atom blows ARM clean out of the water. Those who somehow have acquired a different perception, do not understand how little various devices ask of the ARM, compared to a PC.
The situation today, is that Atom, as a processor is various multiples faster, depending upon tasks. That Atom also has the advantage on performance per SI area, and performance per $. ARM does have two advantages though. It has higher performance per Watt, somewhat. But the biggest advantage comes when we consider the total package. ARM is typically rolled into its own tailormade environment in a SOC. This is where ARM makes its huge gains, And Atom, sofar, starts to look disappointing, thanks to crappy, powerhungry chipsets. I would expect Intel to come up with ways to nullify this advantage though, i.e. more integration.
Rounding off, I would suggest that if we also consider power consumption from antennas, audio, LCD/LED displays, then the significance of ARM's performance per Watt is greatly reduced. Finally, as more and more is asked from mobile devices, x86 is going to be increasingly favored.
So the potential of Intel's position is good. It's just that I think a lot of players would rather use their own SOC with a licenced ARM, than just become another low-profit, competing manufacturer for Intel.

Sorry, let me re-phrase one line: It (ARM) has higher performance per Watt. period. Forget that "somewhat", because the difference is in relative terms quite considerable. If it's so in significant terms remains to be seen.