[AT] Intel trying to release 7 nm in 2021

[TheGiant]

Intel said its Xe Datacenter GPU that’s powering the Aurora 21 Exascale Super computer

ok, maybe something that doesn't require balcony nuclear reactor to run

I remember reading something from a Intel manageris woman that 10nm tech will be first ready for server...10nm doesn't look that way, maybe 7nm will?

 

[IntelUser2000]

I remember reading something from a Intel manageris woman that 10nm tech will be first ready for server...10nm doesn't look that way, maybe 7nm will?

"Datacenter GPU" from Dayman's post you quoted is server.

 

[beginner99]

I mean intel ahd to be doing something in the last half a decade with lakes and 14nm. Entirely possible that in 2021 we will get another Conroe-Moment if all things line up (7nm EUV and all their packaging stuff).

 

[Ajay]

From the OP linked article:

While Intel's first 7 nm product will be launched in 2021, Intel stresses that high-volume manufacturing (HVM) using the technology will begin in 2022 when the technology will be used not only for a server GPU, but also a server CPU. So, expect more 7 nm products three or four years down the road.

SAD!

No really useful details here, like any metric on how 7nm looks. Just another dog & pony show with Intel's 'innovative' blue themed slide deck.

 

[NostaSeronx]

No really useful details here, like any metric on how 7nm looks.

That will be later...
NXE:3400C launches later this year. By the time, 2021-2022 comes by Tput upgrade will push it up from 145~155 wph to 175~185 wph. (EUV)

NZ3C is set for 2020 at >100 wph with a gradual throughput increase to >200 wph by 2022. (JFIL)

Intel also has Tilted Ion Implantation. (TII-based double patterning, higher accuracy/throughput and lower cost than SADP/SAQP)

 

[jpiniero]

From the OP linked article:
SAD!

No really useful details here, like any metric on how 7nm looks. Just another dog & pony show with Intel's 'innovative' blue themed slide deck.

Sounds about right. They probably will lose the Aurora contract (to AMD?) if they don't deliver in 2021. So as soon as possible they will try to produce the chiplets no matter the yield.

 

[Topweasel]

That's cute and all, but you're defending someone saying "How the mighty have fallen", to a company who owns 90% of the overall market.

It's not about being in denial. It's about silly statements. The same ones that were made as I previously noted. All the fans were out in force saying Intel was going to crumble.

Do you know what the word cyclical means? The things come in waves. So AMD takes a bit of market share for the next year and half. What happens with Intel rolls out the 7nm? They are right back in the game and they have the revenue to create very attractive pricing to manufactures and vendors. AMD has a good product and road map, but saying Intel has fallen greatly is much an exaggeration. I also never implied or said they weren't going to gain some market share.

Right back in the game they shouldn't have left. Them not all of a sudden not selling any CPU's isn't what it takes for the "Mighty has fallen". There are a dozen things Intel is expected to be the best at by a considerable margin and they either aren't or have such a short lead that it would be considered parity. That isn't the Intel that supposedly learned from their mistakes in 2006-2008. That isn't what Intel has Historically been. It doesn't take seeing their market share to tumble for those in the know to recognize that Intel has been backsliding into their Netburst/Itanium like days. It doesn't have to even require AMD to make the pushes that they are making now. The fact that TSMC leapfrogged them is huge, the fact that they have missed a proccess window let alone for 3+ years is incredible. Maybe Intel can course correct with 7nm. But then it's close to TSMC being fully ready with 5nm. We are getting back to the point that performance will be less about process and more about CPU design and that is another avenue that Intel is soooo behind on. They haven't done anything significant in almost 8 years.

 

[Ajay]

Sounds about right. They probably will lose the Aurora contract (to AMD?) if they don't deliver in 2021. So as soon as possible they will try to produce the chiplets no matter the yield.

Fortunately for Intel, compute oriented GP-GPU is more easily partitioned than graphics workloads.

 

[LikeLinus]

Right back in the game they shouldn't have left. Them not all of a sudden not selling any CPU's isn't what it takes for the "Mighty has fallen". There are a dozen things Intel is expected to be the best at by a considerable margin and they either aren't or have such a short lead that it would be considered parity. That isn't the Intel that supposedly learned from their mistakes in 2006-2008. That isn't what Intel has Historically been. It doesn't take seeing their market share to tumble for those in the know to recognize that Intel has been backsliding into their Netburst/Itanium like days. It doesn't have to even require AMD to make the pushes that they are making now. The fact that TSMC leapfrogged them is huge, the fact that they have missed a proccess window let alone for 3+ years is incredible. Maybe Intel can course correct with 7nm. But then it's close to TSMC being fully ready with 5nm. We are getting back to the point that performance will be less about process and more about CPU design and that is another avenue that Intel is soooo behind on. They haven't done anything significant in almost 8 years.

Still, a stupid argument that lacks actual data. The main point is that 90% of the end users DO NOT CARE or even know what 10nm or 7nm means. You keep making these statements that Intel hasn't innovated for 8 years, for over 3 years TSMC is doing all these incredible things and Intel is just stagnate.

But you fail to realize that Intel still owns 90% of the over all market, despite all these "downfalls" you claim. Because, be brutally honest with yourself, no damn normal person has a clue how many nanometer their chip is, lol. Hell, I don't even know. We've for Ryzen 7 2700X in my son's gaming computer and I have an Intel 9700K. There's virtually no difference. I chose Intel because I do video production and it's a far superior chip for rendering and AE. We chose Ryzen for my son's gaming computer. We don't actually care about the die size or manufacturing process. If it performs, it performs.

It's just fanboys trying to act like "their" company is better. It's pretty stupid, if you ask me. But Intel hasn't lost a tremendous amount of market share and they will catch back up.

 

[maddie]

Still, a stupid argument that lacks actual data. The main point is that 90% of the end users DO NOT CARE or even know what 10nm or 7nm means. You keep making these statements that Intel hasn't innovated for 8 years, for over 3 years TSMC is doing all these incredible things and Intel is just stagnate.

But you fail to realize that Intel still owns 90% of the over all market, despite all these "downfalls" you claim. Because, be brutally honest with yourself, no damn normal person has a clue how many nanometer their chip is, lol. Hell, I don't even know. We've for Ryzen 7 2700X in my son's gaming computer and I have an Intel 9700K. There's virtually no difference. I chose Intel because I do video production and it's a far superior chip for rendering and AE. We chose Ryzen for my son's gaming computer. We don't actually care about the die size or manufacturing process. If it performs, it performs.

It's just fanboys trying to act like "their" company is better. It's pretty stupid, if you ask me. But Intel hasn't lost a tremendous amount of market share and they will catch back up.

What these fora try to do is predict and speculate on the future. Don't try to be the guy claiming to be flying on the way to hitting the ground.

 

[Ajay]

What these fora try to do is predict and speculate on the future. Don't try to be the guy claiming to be flying on the way to hitting the ground.

While that has always been the case, it’s also true that we used to deal in more facts too. With companies being more tight lipped than they used to be, and fewer insiders posting on these boards, the speculation rages on harder than ever.

 

[dmens]

If they are indeed sampling 10nm server parts (carefully didnt say core count lol)

That would be absolute BS. This 1H'20 (read: late June) shipment will be a trickle of thin-and-light crippled Icelake laptop parts and Cannonlake disappears down the memory hole. You will not see 10nm server parts for a very long time.

I do like how the slide brags about power-on with customers. The issue isn't functionality, it is manufacturing. Typical senior corporate obfuscation from the Santa Clara corruptocrats.

 

[TheGiant]

That would be absolute BS. This 1H'20 (read: late June) shipment will be a trickle of thin-and-light crippled Icelake laptop parts and Cannonlake disappears down the memory hole. You will not see 10nm server parts for a very long time.

I do like how the slide brags about power-on with customers. The issue isn't functionality, it is manufacturing. Typical senior corporate obfuscation from the Santa Clara corruptocrats.

you have more info than we it seems !

care to share with us ?

 

[Zucker2k]

I love the title of this thread. "Trying" is definitely an apt characterization for all roadmaps going forward.

On topic: I think after 4 years of Skylake it's well past the time for Intel to release something totally new. Lots of skepticism from the usual suspects but Intel has been consistently insistent on it's roadmap, and now this. Things are looking up indeed.

 

[TheELF]

There are a dozen things Intel is expected to be the best at by a considerable margin and they either aren't or have such a short lead that it would be considered parity.

What are you even talking about?
Intel can clock 20% higher which makes them at least 20% faster in anything that can utilize enough of the CPU(which would be the only reason to even buy a CPU like that) ,looking at benchmarks intel is 20-30% faster in non AVX workloads,that's not only enough of a difference to choose one over the other this is enough of a difference to even upgrade from one to the other.
In AVX workloads the difference is 90% + which is not even in the same ballpark it's a completely different league intel is playing in.

The fact that TSMC leapfrogged them is huge, the fact that they have missed a proccess window let alone for 3+ years is incredible. Maybe Intel can course correct with 7nm. But then it's close to TSMC being fully ready with 5nm. We are getting back to the point that performance will be less about process and more about CPU design and that is another avenue that Intel is soooo behind on. They haven't done anything significant in almost 8 years.

nm means nothing if it's not supported by the clocks, nobody is going to buy 7nm or 5nm if they are clocked at 4Ghz.
Yes intel would be pretty stupid if they released a 10nm chip at 4Ghz or even 4,5Ghz ,why on earth would they release a 10nm chip that is slower then their current one?

 

[coercitiv]

nm means nothing if it's not supported by the clocks

this summer will be hilarious

 

[Zucker2k]

What are you even talking about?
Intel can clock 20% higher which makes them at least 20% faster in anything that can utilize enough of the CPU(which would be the only reason to even buy a CPU like that) ,looking at benchmarks intel is 20-30% faster in non AVX workloads,that's not only enough of a difference to choose one over the other this is enough of a difference to even upgrade from one to the other.
In AVX workloads the difference is 90% + which is not even in the same ballpark it's a completely different league intel is playing in.

nm means nothing if it's not supported by the clocks, nobody is going to buy 7nm or 5nm if they are clocked at 4Ghz.
Yes intel would be pretty stupid if they released a 10nm chip at 4Ghz or even 4,5Ghz ,why on earth would they release a 10nm chip that is slower then their current one?

If Zen 2 comes in clocking less than Zen+ it'll put things in perspective for me, at least, why Intel has sat on 10nm for so long. Intel was the leader of the pack until 10nm and was bound to run first into the unique challenges of a node shrink at 10nm and below with their especial focus on density. Moreover, Intel being in a unique position of foundry and chipmaker could immediately assess the comparative disadvantages of their new node/arch compared to the previous (read legendary 14nm++) iteration and do the math in situ. AMD doesn't have that luxury. They don't have many options. They either go with TSMC 7nm now, or wait for the respin. We shall soon see a clearer picture of the various factors in play once the chips are released.

 

[exquisitechar]

Right back in the game they shouldn't have left. Them not all of a sudden not selling any CPU's isn't what it takes for the "Mighty has fallen". There are a dozen things Intel is expected to be the best at by a considerable margin and they either aren't or have such a short lead that it would be considered parity. That isn't the Intel that supposedly learned from their mistakes in 2006-2008. That isn't what Intel has Historically been. It doesn't take seeing their market share to tumble for those in the know to recognize that Intel has been backsliding into their Netburst/Itanium like days. It doesn't have to even require AMD to make the pushes that they are making now. The fact that TSMC leapfrogged them is huge, the fact that they have missed a proccess window let alone for 3+ years is incredible. Maybe Intel can course correct with 7nm. But then it's close to TSMC being fully ready with 5nm. We are getting back to the point that performance will be less about process and more about CPU design and that is another avenue that Intel is soooo behind on. They haven't done anything significant in almost 8 years.

Intel has done plenty in the past 8 years, it's just been held back because of the 10nm issues. Their processor architects haven't been on vacation this whole time, the new stuff they've been working on will be rolled out rapidly. Don't underestimate their "CPU design" just because they've been stuck on Skylake for years.

 

[DrMrLordX]

This 1H'20 (read: late June) shipment will be a trickle of thin-and-light crippled Icelake laptop parts

I thought those parts were launching this year, not 2020. Unless they're just going to launch more of them as a refresh (yay).

nm means nothing if it's not supported by the clocks, nobody is going to buy 7nm or 5nm if they are clocked at 4Ghz.

. . . they will if they get twice as many cores and higher IPC. The way you're talking, it's like nobody bought an 1800x. Or a 1700.

Yes intel would be pretty stupid if they released a 10nm chip at 4Ghz or even 4,5Ghz ,why on earth would they release a 10nm chip that is slower then their current one?

14nm ought to be enough for anyone.

If Zen 2 comes in clocking less than Zen+ it'll put things in perspective for me

There's already a 16c/32t ES floating around with a boost clock of 4.2 GHz. An ES, mind you. AMD engineering samples tend to run far below launch clocks. I do not think you will have to worry about any new perspectives.

 

[Zucker2k]

There's already a 16c/32t ES floating around with a boost clock of 4.2 GHz. An ES, mind you. AMD engineering samples tend to run far below launch clocks. I do not think you will have to worry about any new perspectives.

I hope not. Hehe.

But bear in mind that the bar has been significantly higher for Intel.

 

[Markfw]

What are you even talking about?
Intel can clock 20% higher which makes them at least 20% faster in anything that can utilize enough of the CPU(which would be the only reason to even buy a CPU like that) ,looking at benchmarks intel is 20-30% faster in non AVX workloads,that's not only enough of a difference to choose one over the other this is enough of a difference to even upgrade from one to the other.
In AVX workloads the difference is 90% + which is not even in the same ballpark it's a completely different league intel is playing in.

nm means nothing if it's not supported by the clocks, nobody is going to buy 7nm or 5nm if they are clocked at 4Ghz.
Yes intel would be pretty stupid if they released a 10nm chip at 4Ghz or even 4,5Ghz ,why on earth would they release a 10nm chip that is slower then their current one?

I would say what are YOU talking about ? Care to back that up with some benchmarks from a reputable site ?

 

[TheELF]

. . . they will if they get twice as many cores and higher IPC. The way you're talking, it's like nobody bought an 1800x. Or a 1700.

Yes exactly,they bought them because of the number of threads not because of smaller nm than intel or because of more IPC than intel.

I would say what are YOU talking about ? Care to back that up with some benchmarks from a reputable site ?

Which part? that you can clock a 9900k to 5Ghz and ryzen tops out at 4.2?

 

[Markfw]

Yes exactly,they bought them because of the number of threads not because of smaller nm than intel or because of more IPC than intel.

Which part? that you can clock a 9900k to 5Ghz and ryzen tops out at 4.2?

I wanted benchmarks, not BS. But just to show you where Anandtech stands, from their review of the 9900k "AMD’s R7 2700X is very competitive in almost every test, while they might not be the best, they’re more cost-effective. "

Thats not 20-30% Thats maybe <5%, and in that review the 2700x wins several times.

 

[TheGiant]

https://www.techspot.com/review/1744-core-i9-9900k-round-two/

here we have both versions of the 9900K and it is not 5%, not even close..

this round with 14nm Intel won by a mile, but that is not denying the succes AMD did with their 2700X, which deserves a medal definitely

but here about 7nm

please anyone can tell the real comparison between the TSMC 7nm (AMD, apple...) and Intel's 7nm ?

thanks

 

[Markfw]

https://www.techspot.com/review/1744-core-i9-9900k-round-two/

here we have both versions of the 9900K and it is not 5%, not even close..

this round with 14nm Intel won by a mile, but that is not denying the succes AMD did with their 2700X, which deserves a medal definitely

but here about 7nm

please anyone can tell the real comparison between the TSMC 7nm (AMD, apple...) and Intel's 7nm ?

thanks

Looks to me like they trade blows. The 2700x wins the very first benchmark. It looks like its between the 95 watt and the unlimited benches on average.

And from your own link, in the conclusion:
It’s a $500 8-core desktop CPU competing with a $300 8-core desktop CPU. As we just saw with the 95-watt limit, it’s barely any faster than the Ryzen 7 2700X. In fact, in some tests it’s slower, and that’s an awful result for a CPU that costs ~70% more. "