Intel details advances to make upcoming chips faster, less costly

By now, you likely know the story: Intel made major manufacturing missteps over the past several years, giving rivals like AMD a major advantage, and now the x86 giant is in the midst of an ambitious five-year plan to regain its chip-making mojo. This week, Intel is expected to detail just how it's going to make chips in the …

[ShadowSystems]

Wake me when you're relevant again.

Everyone else has been eating your lunch, stealing your lunch money, kicking sand in your face, & forcing you to sit at the toddler table for so long that you can make all the bluster you want, it won't make a damned bit of difference until you regularly, reliably, & repeatedly deliver technology that's not just "as good as" but *better than* everyone else's last year's kit.

You can rebrand your stuff to be "UltraSpiffySubMesonSuperDuperElectronMicroscopic", it doesn't mean a thing if you can't prove the bluster to be anything more than the dying tantrum of a spoiled brat that's long since run out of toys to fling from the pram. You're akin to a Howler Monkey screaching at the competition & flinging shite in the hopes someone, anyone will be distracted by the shite show to confuse the fact that you're not the powerhouse you used to be.

TL;DR: Put Up or STFU, the bluster & marketing spin is worth SFA, much like you.

[Andy 73]

Re: Wake me when you're relevant again.

Point on the diagram to where Intel hurt you.

It's a company, not an ex girlfriend - possibly not worth this level of anger. If they can deliver, it's a benefit to everyone (competition is good). If not, sucks to be them, but there remain plenty of alternatives. From an outsider's perspective, it's interesting to hear they've at least acknowledged their previous approach was a dead end.

[sreynolds]

Re: Wake me when you're relevant again.

Point on the diagram to where Intel hurt you.

Don't you mean on this anatomically accurate doll?

[jngreenlee]

Re: Wake me when you're relevant again.

Sir, you mean "on this process leadership roadmap", I suspect

[Arthur Daily]

Re: Wake me when you're relevant again.

All bluster. What are you going to do, to beat TSMC. Oh I see, this and that, and hope to come in 3rd. Did they say they would be able to at least match TSMC: Nope. I also see speculative execution flaws - do not appear to be fixed yet. The good news is China is free to ramp up its laggard tech and fabs. May the best player win.

[Pascal Monett]

The best player will win.

This is tech.

If China comes out with a new x86 design that is blisteringly fast and efficient, that's what people are going to buy - politics be damned.

[Yet Another Anonymous coward]

>China comes out with a new x86 design

The ultimate death of communism

[OhForF']

"The best player will win. This is tech."

Unfortunately that doesn't mean the best technology will win.

Offering the best techonology doesn't automatically make you the best (most sucessful) player.

[pavel.petrman]

Ambitious five year plan...

I hope it works out better for Intel than it used to for those countries which, too, bet their future on one ambitious five year plan after another.

[Anonymous Coward]

Re: Ambitious five year plan...

All the chip manufacturers have ambitious five year plans. They're called Roadmaps. ElReg has been reporting on the new ones for the last week or so.

We should all wish all of them success because, as Daft Punk said, "Work it harder, make it better

Do it faster, makes us stronger."

They are engineering plans and comparing them to the economic five year plans of countries is disingenuous.

[Yet Another Anonymous coward]

Re: Ambitious five year plan...

>They are engineering plans

Of for sure, we asked a bunch of engineers how long it would take to implement a new EUV methodology that nobody has attempted before and their critical path model predicts a timeline which exactly aligns with our corporate financial reporting period and the outstanding during of our current corporate bonds

[imanidiot]

picking nits

"In practice, this means Intel can reduce the number of layers required to etch chip designs onto silicon wafers from five to one."

More accurately, not needing multi-patterning requires the number of exposures/masks required. With multi patterning those 5 exposures would still be on the same 'layer'/z-height on the wafer.

[imanidiot]

Re: picking nits

Correction:

More accurately, not needing multi-patterning reduces the number of exposures/masks required.

Because of course proofreading is just so damn hard or something...

[Anonymous Coward]

Re: picking nits

It's funny: I hardly understood the technical details you presented, but in attempting to understand what you said, I actually knew you meant "reduced" instead of "required" before you made your correction. Amazing how the human brain processes information it barely understands. *chuckles*

[confused and dazed]

Good

All sounds very sensible and logical to me. Yes they have milked their leadership in the past, but we need this competition

[Wade Burchette]

Re: Good

I hope Intel can deliver. Competition is a good thing because it means lower prices for us. I saw what AMD before Intel's Alder Lake 12th gen line came out.

So I hope Intel can deliver. But I am not confident than they can. Where is Intel's discrete GPU? Here it is the middle of June and there is still no hint of an actual release date. It is looking more and more like that, instead of competing against current generation from Nvidia and AMD, it will be competing against last generation. It also appears that Intel's 13th gen of CPU's are also delayed. Once again, Intel could be finding themselves competing against a last gen product.

AMD has been meeting their roadmaps. The only thing that delayed them was COVID, something out of their control. Intel needs to focus on first delivering on time first. What good is the announcement about how good their '4' fab line is if I cannot buy a product with it?

(Oh, and please please please Intel, make your socket last more than two different lines. It would be nice if, like AMD, I can upgrade my processor to a processor that came out 4 years after the motherboard was released.)

[steelpillow]

capacitors on speed

"Increased capacitance results in fewer large voltage swings, which, in turn, increases the available voltage to the CPU and allows it to run at a higher frequency"

That is a strange way of putting it; presumably it is referring to power rail smoothing capacitors. But that only allows an increase in frequency to be obtained by driving a given circuit nearer its limits, where it is less efficient and consumes more power.

So you have to find bigger power savings elsewhere and then trade off some of that for speed.

[Anonymous Coward]

Re: capacitors on speed

There is a decent power saving coming from the die shrink (e.g. this is why AMD are still ahead on power efficiency - they are a node ahead on fabrication)

[Bartholomew]

Re: capacitors on speed

> That is a strange way of putting it; presumably it is referring to power rail smoothing capacitors.

Every transistor acts as a capacitor, it is one of the main reasons it takes time to switch them on (and off).

Making transistors smaller reduces the charge/discharge time. So they could in theory switch faster at a lower voltage. But the downside is that you may have more heat to dissipate per unit area, if you switch too fast, which could melt the chip! It is the main reason clock rates are not going to increase by orders of magnitude on silicon at least unless we have some new improved cooling technology.

[steelpillow]

Re: capacitors on speed

"Making transistors smaller reduces the charge/discharge time. So they could in theory switch faster"

Indeed. However the claimed advance is to make capacitors larger. That is hardly consistent with your observation. Hence my assumption that rather different capacitors are involved.

[Sil]

What a joke. If everything goes as planned, Intel will offer 7 nm when competitors go 4 nm.

Surefire way to regain leadership.

[Dave K]

Yes, yes, but branding!

"It's 7nm, but we consider it the same as everyone else's 4nm, so we call it Intel 4". Genius!

Of course the proof will be in the pudding when the performance and power requirements of "Intel 4" chips are compared with those from genuine 4nm fabrication processes...

[Peter2]

It's worse than that.

https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_3nm

TSMC’s 3nm technology (N3) will be another full node stride from our 5nm technology (N5), and offer the most advanced foundry technology in both PPA and transistor technology when it is introduced. N3 technology will offer up to 70% logic density gain, up to 15% speed improvement at the same power and up to 30% power reduction at the same speed as compared with N5 technology. N3 technology development is on track with good progress. N3 technology will offer complete platform support for both mobile and HPC applications, which is expected to receive multiple customer product tape-outs in 2021. In addition, volume production is targeted in second half of 2022.

So AMD is potentially capable of deploying 3nm chips from second half 2022; ie within the next few months, well before Intel deploys their 7nm process. Personally, i'd hold those back and deploy them in volume the day that Intel produces a comparable product to the existing lineup.

Either Intel gets their processes working, or they are going to be forced to go to TSMC so they are able to compete, which would mean competing on chip design instead of manufacturing process.

That's be disastrous for Intel; they have a terrible track record of managing that since the days when the AM386 chips turned out the same level of performance as the Intel 486.

[GFK1]

Anybody claiming AMD 3nm is comparable to Intel 7nm hasn't been paying attention at all, or is being deliberately disingenuous.

Those numbers long since ceased to be comparable metrics.

[Peter2]

Ok, they aren't comparable metrics.

This is; Intel is incapable of matching the deployment of new processes that TSMC is routinely rolling out.

Even if you buy the "Intel 4" marketing of their 7nm process "Intel 4" is going to be available in sample levels after TSMC's 3nm process is available in volume according to the roadmaps of both. By the time Intel is deploying that in volume, TSMC and their customers like AMD will have already deployed their 3nm enhanced.

TSMC has a good track record of making conservative guesses on their roadmap and delivering early for their customers; Intel has a good track record of over promising and delivering late.

By the time "Intel 3" which is supposed to counter TSMC's 3nm process is available in any quantity then TSMC is going to be rolling out their 2nm process. And Intel 3 is going to be equivalent to TSMC's 3nm; not 3nmE which will have then been in volume production for a year and a half at that point.

According to Intel's roadmap, they aren't going to be deploying anything equivalent to TSMC's processes in the next 5 years.

[imanidiot]

What GFK1 said. Intels way of measuring feature/node size and AMDs (really TSMCs) way of measuring node size are not the same. Usually there was a step between them though, so Intel 7nm would be comparable to TSMC 5 nm. If that still holds TSMC will still be a step ahead of them. We'll see how it works out. Both Intel and TSMC have been figuring out how to properly utilize EUV litho systems/processes in their lines and it's been a struggle, from what I understand especially for Intel.

[Anonymous Coward]

Performance improvement

So have we gotten back the performance we lost to the hack-arounds for Spectre/Meltdown/etc? Or is that loss now just automatically included in the calculations and nobody worries about it anymore?

[Ace2]

Missed opportunity

Ben Sell didn’t go into sales & marketing?

[Benegesserict Cumbersomberbatch]

Re: Missed opportunity

It's not his name, it's his stock position.

[localzuk]

Believe it when I see it

Intel have made so many promises over the years and have simply not delivered. We'll see if this is more of the same, or if AMD will keep their lead.

[Anonymous Coward]

What is the mass market that needs the extra performance? On the other hand longer battery charge life does sound useful.