Intel Says It Will Move Away From 'Tick-Tock' Development Cycle

An anonymous reader writes: In its latest annual report, Intel says that it will be moving away from its decade-old "tick-tock" strategy (PDF) for developing new chips. From the company's 10-K filing, "We expect to lengthen the amount of time we will utilize our 14nm and our next generation 10nm process technologies, further optimizing our products and process technologies while meeting the yearly market cadence for product introductions." Anand Tech's Ian Cutress explains, "Intel's Tick-Tock strategy has been the bedrock of their microprocessor dominance of the last decade. Throughout the tenure, every other year Intel would upgrade their fabrication plants to be able to produce processors with a smaller feature set, improving die area, power consumption, and slight optimizations of the microarchitecture, and in the years between the upgrades would launch a new set of processors based on a wholly new (sometimes paradigm shifting) microarchitecture for large performance upgrades. However, due to the difficulty of implementing a 'tick', the ever decreasing process node size and complexity therein, as reported previously with 14nm and the introduction of Kaby Lake, Intel's latest filing would suggest that 10nm will follow a similar pattern as 14nm by introducing a third stage to the cadence."

Comments

R.I.P. Andy Grove

[Thud457]

My grandfather's clock was too large for the shelf,
So it stood ninety years on the floor;
It was taller by half than the old man himself,
Though it weighed not a pennyweight more.
It was bought on the morn of the day that he was born,
And was always his treasure and pride;
But it stopped short â" never to go again â"
When the old man died.

Ninety years without slumbering
(tick, tock, tick, tock),
His life's seconds numbering,
(tick, tock, tick, tock),
It stopped short â" never to go again â"
When the old man died.

Moore's law, say hello to the law of economics

[JoeyRox]

At $5B+ for a single fab and the market for computers continuing its backward slide it's no surprise that Intel is putting the brakes on its capital expenditures.

Re:Moore's law, say hello to the law of economics

[Anonymous Coward]

I also wonder what they call paradigm shifting micro-architectures. Basically, apart from the PIV/Netbu(r)st series, all Intel processors are descendants of the Pentium-Pro, their first OOO processor. The changes have been in the area of completely different FPU units (good riddance for the x87 stack), switching to 64 bit (you have to thank AMD for that), and a few other improvements. But putting the memory controller or the GPU on the chip, using faster/wider I/O busses, or multiplying the number of cores is _not_ a paradigm shift, it's only using the transistor budget wisely because you can't design a core which needs 1 billion transistors (even if most of these are in the cache).
Bottom line, the core (no pun intended) of their current processors is closer to the 20 year old Pentium-Pro than to the Pentium IV or the original Pentium (which itself was largely a dual 486). Now I don't criticize the decision: if it isn't broke, don't fix it (Netbust was an ill-fated attempt). What I object to is calling it a paradigm shift when there have been only incremental improvements (in the micro-architecture) between PPro, PII, PIII, and the whole series of Core processors.

Re:Moore's law, say hello to the law of economics

[theendlessnow]

And there simply is no competition anymore. Tick-tock was designed to hammer, hammer and keep on hammering against AMD until they were dead, deAD, DEAD! (for those that don't know AMD used to compete against Intel)

Re:Moore's law, say hello to the law of economics

[castionsosa]

You hit the nail on the head. "Good enough" has knocked Moore's Law off the rails. Since there isn't that much demand, other than adding cores for virtualization [1], it isn't surprising that Intel is backing off the gas pedal with CPU development.

There are other things as well to add to a CPU. Disk I/O hasn't kept up with capacity gains, and there is always working on better power management which is something I'm sure Intel's enterprise customers are heavily damanding for PR reasons.

[1]: The ideal would be faster cores, since Microsoft has hopped on the Oracle and Sybase bandwagon and started licensing by core, and not CPU socket, but more cores is better than nothing.

Re:Moore's law, say hello to the law of economics

[Anonymous Coward]

New instructions don't change the fundamental flow of data inside the processor. In the PPro presentation, they said that the FPU needed 86 bit wide busses (80 bits data + status) and that this was " a lot of bits". Now they have AVX256 and 512 is right around the corner. Using parallelism to implement vector instructions is great for some tasks, but a compiler, for example or an interpreter (Python, Ruby, Perl) still executes mostly basic i386 instructions (or their 64 bit extensions).
Making the instruction set byzantinely complex does not help programmers and compiler writers, and the encoding is so complex that there two instruction caches on some processors: one in the native, inscrutable encoding, and one in another encoding easier on the decoders and schedulers. A saner, simpler, encoding (ARM and MIPS are quite good in this respect) would avoid this dual cache (which in the end means that the first level instruction cache is smaller because of transistor budgets).

Re:Moore's law, say hello to the law of economics

[Kjella]

Actually it's just Moore's law breaking down, the difficulty is in producing smaller transistors, the technology can't keep up. We know Intel had to delay the 14nm launch because of bad yields, now on 10nm it's probably a lot worse. And to go beyond that you need extreme ultraviolet lithography (EUV) which is still heavily in the R&D phase. I'm guessing that what Intel really knows at this point is that with a lot of tweaking they can probably do 10nm with acceptable yields using mostly known technology. What the world really looks like after three generations of 10nm? I don't think anybody knows. Intel once had a roadmap where they were tick-tocking all the way down to 5nm. This new roadmap is also just wishful thinking on where they'd like to be.

Digital computers are reaching the end

[110010001000]

This will make a LOT of people here mad, but the exponential growth computational power of digital computers is ending. We can no longer count of the computers of tomorrow to be significantly faster or have more memory than today. If you have been following the industry closely you can already see start to happen 10 years ago. So we can forget about projections that used the argument of exponential growth creating the "Singularity" or "AI". There just simply won't be enough processor power available with classical digital computers. The computer you use 10 years from now will look and perform a lot like the one you have today.

Re:Digital computers are reaching the end

[basscomm]

This will make a LOT of people here mad, but the exponential growth computational power of digital computers is ending. We can no longer count of the computers of tomorrow to be significantly faster or have more memory than today. If you have been following the industry closely you can already see start to happen 10 years ago. So we can forget about projections that used the argument of exponential growth creating the "Singularity" or "AI". There just simply won't be enough processor power available with classical digital computers. The computer you use 10 years from now will look and perform a lot like the one you have today.

Heck, the computer I use 10 years from now might very well be the same computer that I'm using today.

Re:Digital computers are reaching the end

[peragrin]

What you are forgetting is that to keep processor fabs paid for they keep shrinking everything else. While CPUs are 14nm most gpus are not. Ram is not.

In time expect to see ram, gpus, and the other components shrink as well. In 10 years you will buy a computer where all transistors inside it are at 14nm or less and it is Using a fraction of the power.

Re: Digital computers are reaching the end

[armanox]

They already confirmed they were screwing us by making future processors lower power rather than faster. They're shoving their green agenda down our throats.

Oh yes, how dare they focus on things that effect users, like how long the battery in their laptop lasts, or their electric bills. The majority of users aren't seeing issues with CPU speed, so it is becoming less of a focus then other factors. Heaven forbid they focus on the consumer's needs.

Re: Digital computers are reaching the end

[Mal-2]

Oh yes, how dare they focus on things that effect users, like how long the battery in their laptop lasts, or their electric bills. The majority of users aren't seeing issues with CPU speed, so it is becoming less of a focus then other factors. Heaven forbid they focus on the consumer's needs.

If they could keep whipping the "faster, faster, faster" horse, they would. When the primary advantage of a new computer over an old one is that it takes less power and generates less heat, people don't see much pressure to upgrade. The old one still works just fine, even if the cost of operation is higher. This is not to say that pushing "smaller, cooler, quieter" is a bad thing for the world at large. It's obviously good. But it's not as good for Intel as pushing speed at all costs used to be. Therefore the conclusion has to be that they're doing it this way because as successful as the old way was for them, they can't make it work any longer.

Most people wouldn't need the power anyway

[jones_supa]

For an ordinary joe it won't matter that much. Most of the services he uses (Facebook, Twitter, Spotify, Netflix, Skype, lightweight gaming) could be implemented even on a Pentium II with a little bit of optimization. Even Microsoft does not bother artificially bloating their operating system anymore.

Should have mentioned diminishing returns

[avandesande]

Differences in performance (speed, power consumption etc.) are now almost imperceptible between process changes.

Re:Should have mentioned diminishing returns

[avandesande]

What do you mean by puts out? They do that by shrinking the die and creating a new fab, which is precisely what the article is talking about. Back in the day you could see a double or triple in clock speed with each iteration. You really think that Intel is sitting on a 10ghz i8 for business reasons?

Moore's law is dead; physics killed it

[dlenmn]

I'm not picking on you in particular, but I'm seeing a lot of posts implying that Moore's law could keep going but it's too expensive, there's not enough competition to warrant it, etc. The fact is that physics is the nail in the coffin for Moore's law. Making small fab processes is getting more and more difficult because these size scales are super tiny, and the difficulty means that Moore's law simply cannot keep going because we have to develop fundamentally new technology -- not just scaled down current technology.

There's a reason Intel is planning to stop using Silicon at 7 nm (not clear what they'll move to -- maybe indium gallium arsenide), and getting up to production quality with a new material is a huge task that is fundamentally incompatible with Moore's law. (InGaAs is not "new" per se, but InGaAs has never seen real commercial use; it has been confined to research labs.)

There's also a reason that research in classical (not only quantum) computing with superconducting circuits is again being seriously researched by commercial enterprises -- including companies like Northrup Grumman which are not traditionally associated with designing computer chips. (IBM poured a lot of money into superconducting computers in the 1980s but ultimately gave up because Si computing was marching along just fine. I think that IBM is back in the superconducting game too.) Again, getting superconducting circuits up and running is _hard_ and fundamentally incompatible with Moore's law.

Moore's law is intrinsically dead. End of story. Even if/when the non-Si chips get up and running, I don't expect that Moore's law will be revived. 7 nm equates to about 14 silicon atoms. The end of the road is in sight. It's trying to march through quicksand from here on out.

PS. I don't get the "lack of competition" hypothesis for why Intel is slowing down; there are a number of manufacturers matching or closing in on Intel's fab process. E.g. Samsung and Globalfoundries are already at 14 nm. TSMC is at 16 nm. These aren't in direct competition with Intel at the moment, but they will be if Intel ever gets serious about putting their chips in things other than desktops/laptops/servers. Intel isn't stupid; they see these other companies as competitors, and Intel really wants a leg up on them. If Intel could keep up with Moore's law, they would.

Re:new cycle

[ChrisMaple]

Hickory Dickory Dock.

Moving to a three stage process?

[Chas]

I vote that we call it "Boom Shaka Laka"!

ARM only replacing x86 for servers and consumers

[raymorris]

Yeah, ARM can never replace x86. There will never be a day when consumers mostly buy ARM devices rather than x86. Well not until 2014 anyway. The fact is, most CPUs purchased in the last two years were ARM.

In the datacenter, power (and it's associated cooling) is expensive, so we're already starting to see ARM replacing x86 in the datacenter too.

  Business desktops still mostly run x86, because they mostly run Windows and Windows is currently x86-centric. Microsoft has already released an ARM version of Windows, though, and they are currently making a big push toward "apps" that cpu-architecture independent. That is to say, any application written according to Microsoft's recommendations will be ARM compatible.

ARM hasn't completely replaced x86, but ARM does now have most of the market, and the most significant hurdle"for ARM, Windows compatibility, appears to be going away.

Tick/Tock/Wow, new capacitors repost

[justthinkit]

Moore's rule of thumb expired two years ago.

It can't continue forever. The nature of exponentials is that you push them out and eventually disaster happens.
- Gordon Moore, in 2005

Re:Tick-Tock-Tock

[SeaFox]

That's how Intel will waltz into the next decade.

Bedrock?

[Tough Love]

Oh, tick-tock was the bedrock of Intel's success? Silly me, I thought it was more about monopoly control and cutting off AMD's air supply.