From Intel Earnings: Intel says accelerating 10nm product transition, 7nm product transition delayed

staknhalo

Supreme [H]ardness
Joined
Jun 11, 2007
Messages
6,924
Intel said it is accelerating its transition to 10nm products this year with increasing volumes and strong demand for an expanding line up. This includes a growing portfolio of 10nm-based Intel Core processors with "Tiger Lake" launching soon, and the first 10nm-based server CPU "Ice Lake," which remains planned for the end of this year. In the second half of 2021, Intel expects to deliver a new line of client CPU's (code-named "Alder Lake"), which will include its first 10nm-based desktop CPU, and a new 10nm-based server CPU (code-named "Sapphire Rapids"). "The company's 7nm-based CPU product timing is shifting approximately six months relative to prior expectations. The primary driver is the yield of Intel's 7nm process, which based on recent data, is now trending approximately twelve months behind the company's internal target," Intel said.


https://thefly.com/landingPageNews....duct-transition-nm-product-transition-delayed
 
Told you, Intel is all 10nm untill 2022. And at the rate it took them to fix 10nm, I'd expect 7nm desktop/server processors no earlier than 2025.

Intel is slowly bleeding-out, with no signs of saving them. Luckily, Tiger Lake cores are at least faster than Zen 3, but we're going to have to wait until next year to see 10nm versions of those on the desktop. Rocket Lake is a cluster-in-progress, and we're not even sure which process node it is or core architecture...it's also not going to be out before 2021.
 
Last edited:
I'm not planning to build a new system until 2022/23, and figured it was too soon to try and predict whose CPU I'd be using; but if Intel's going to repeat the 10nm debacle at 7nm I'm almost certain I know the answer now.

F
Looking at what Intel is doing lately and listening to their Earnings Conferense, it feels like they are going not to tied their architectural change to the process nodes anymore due to the debacle with their 10nm process. Though I am a bit surprise Intel doesn't outsource part of its foundry to TSMC just to screw AMD.
 
Accelerating???? They were supposed to be transitioned like 4+ years ago. 7nm delays and issues is music to AMDs ears and will have more OEM and large contracts head their way. Intel really needs to get themselves together, it's gone from silly to rediculous to just plain stupid really quickly.
 
Told you, Intel is all 10nm untill 2022. And at the rate it took them to fix 10nm, I'd expect 7nm desktop/server processors no earlier than 2025.

Intel is slowly bleeding-out, with no signs of saving them. Luckily, Tiger Lake cores are at least faster than Zen 3, but we're going to have to wait until next year to see 10nm versions of those on the desktop. Rocket Lake is a cluster-in-progress, and we're not even sure which process node it is or core architecture...it's also not going to be out before 2021.
You got benchmarks of tiger lake vs zen3 already?? Share?
 
Looking at what Intel is doing lately and listening to their Earnings Conferense, it feels like they are going not to tied their architectural change to the process nodes anymore due to the debacle with their 10nm process. Though I am a bit surprise Intel doesn't outsource part of its foundry to TSMC just to screw AMD.

Intel said they were decoupling designs from specific processes a year or so back.

Even ignoring that porting a chip design to a TSMC process would be orders of magnitude harder than changing from one internal node to a second, that would require TSMC to be willing to screw its own long term customers to help a struggling rival. No way. No how.
 
No, but they released Tiger Lake Geekbench results and it's 35% faster per-core over Zen 2 APU. Even an idiot can concede Zen 3 will not be closing that big a gap.

https://www.techpowerup.com/269453/...-ryzen-7-4700u-renoir-in-most-geekbench-tests
Well, I guess I'll be your idiot ;). Oh, you mean the benchmark where the tiger lake was clocked 40% faster than the zen2 chip and managed 35% more performance.... Yeah, I'm an idiot for thinking AMD could come close, lol.
I guess if AMD reduces their thread count and lowers their clock speeds I could agree with you. Did you even read the article or look at the 1 # and just blindly see Intel was unchatchably faster?
So if AMD gets a relatively tame by IPC increase of 10%, they wouldn't need much in the way of frequency increases to "catch up" to tiger lake. Of course, I'm not one to take a single benchmark and make a crazy statement like calling people idiots with a single benchmark that didn't even prove my point. I'll just wait for them to come out and look at real benchmarks, I feel you may be surprised they aren't going to get beat down as badly as you think.
 
Well yeah, we're assuming that the chips were running at the same 15w TDP, installed in similarly-cooled notbooks If that is horseshit, then this comparison is horseshit.

But for now, it's all we have to go on.

Clock speed differences are not cheating in a notebook - if you're running at the same power level, and can clock faster, then you have the more efficient platform.
 
No, but they released Tiger Lake Geekbench results and it's 35% faster per-core over Zen 2 APU. Even an idiot can concede Zen 3 will not be closing that big a gap.

https://www.techpowerup.com/269453/...-ryzen-7-4700u-renoir-in-most-geekbench-tests
The i7 was clocked at 2.8GHz, and the Zen 2 APU was clocked at 2.0GHz - that is a 1.4x higher clock.
If we do the math on that, when the Zen 2 APU is clocked at 2.8GHz, it would score roughly 6892, which is higher than the 6737 score that the i7 scored.

That proves, when both quad-core CPUs are clocked at 2.8GHz, the i7 is roughly 2.3% slower than the Zen 2 APU.
You might want to rethink that "idiot" statement... :meh:
 
Well yeah, we're assuming that the chips were running at the same 15w TDP, installed in similarly-cooled notbooks If that is horseshit, then this comparison is horseshit.

But for now, it's all we have to go on.

Clock speed differences are not cheating in a notebook - if you're running at the same power level, and can clock faster, then you have the more efficient platform.

They're pre-release vendor benchmarks. They should be assumed to be horseshit until such time that proof of being horseshit is found.

And while power/thermal management is a key factor in laptop performance it also gives tons of room to let someone without scruples cook the books, by cranking the power limits up a crazy amount to get a higher score.

And htat's assuming they people who brought us the 28 core 5ghz scam aren't doing something equally sketchy like running their laptop from inside a deep freeze.
 
Looking at what Intel is doing lately and listening to their Earnings Conferense, it feels like they are going not to tied their architectural change to the process nodes anymore due to the debacle with their 10nm process. Though I am a bit surprise Intel doesn't outsource part of its foundry to TSMC just to screw AMD.

Perhaps they tried and got the .... "sorry apple has our fab space reserved till 2023".
 
Perhaps they tried and got the .... "sorry apple has our fab space reserved till 2023".
Could be, in any case, listening to their earnings call, they wouldn't outsource any thing significant until 2023.
 
  • Like
Reactions: ChadD
like this
Zen 3 is going to crush them next year... and the year after Apple may make Intel and AMD cry. At least AMD is probably in a better position to build out some decent ARM + stuff designs if it makes sense.

Intel has a x86 Big.little coming... make sense that all they want to talk about now is features and lifestyle marketing. There going to get destroyed in benches... so now benches are unimportant. Intel is a lifestyle company now.... just sucks for them that the OG lifestyle company is about to upend things this year. I have a feeling Intels big little is going to look like a old school PC clone in comparison to what apple pushes out on the mac pros late next year.

Interesting time to be a tech head... Intel seems to be shitting the bed every night. AMD isn't. And Apple of all companies may actually be the most innovative CPU company. Its like we hit peak bizzaro world.
 
Well at least they have it somewhat working, just high failure rates so they can still meet their deliverable requirements for the supercomputer they agreed to build the chips for at 7nm, so they aren't facing a breech of contract just yet, it just means poor yields (they may take a loss on it) but that is still heaps better than their 10nm debacle. So that's a minor win , but seriously Intel needs to hire somebody from outside to clean them up, they have too much fat, bloat, and likely internal corruption. Their board needs to bring in an outside review team to gut them top to bottom and clean house, Intel is going to get hurt if things continue on this way.
 
Well at least they have it somewhat working, just high failure rates so they can still meet their deliverable requirements for the supercomputer they agreed to build the chips for at 7nm, so they aren't facing a breech of contract just yet, it just means poor yields (they may take a loss on it) but that is still heaps better than their 10nm debacle. So that's a minor win , but seriously Intel needs to hire somebody from outside to clean them up, they have too much fat, bloat, and likely internal corruption. Their board needs to bring in an outside review team to gut them top to bottom and clean house, Intel is going to get hurt if things continue on this way.

They need to start at the top and stop hiring salesmen as CEOs. Time to find the most brilliant engineer still working at Intel... a 30+ year in Intel employee that knows all the engineers that are worth keeping and those who are worth loosing to AMD NV or Apple, and promote them. Axe the bonuses and perks on the folks you want applying for those cool jobs posted at Tesla ect... and get rid of them and let the competition enjoy their skills. AMD went through the same pain... with leadership looking to short term cash in. They wisely got a engineer in charge that knew enough to figure out quickly where the fat was... and who deserved to build the future. It might be time for Intel to do the same... cut some of the fat and focus on making their core products sexy again.
 
They need to start at the top and stop hiring salesmen as CEOs. Time to find the most brilliant engineer still working at Intel... a 30+ year in Intel employee that knows all the engineers that are worth keeping and those who are worth loosing to AMD NV or Apple, and promote them. Axe the bonuses and perks on the folks you want applying for those cool jobs posted at Tesla ect... and get rid of them and let the competition enjoy their skills. AMD went through the same pain... with leadership looking to short term cash in. They wisely got a engineer in charge that knew enough to figure out quickly where the fat was... and who deserved to build the future. It might be time for Intel to do the same... cut some of the fat and focus on making their core products sexy again.
In order for this plan to work, said engineer has to be just as good at management of people and resources. And finding someone like that is incredibly hard. It's generally 'easier' to find a businessman that understands what makes a business powerful and can capitalize on that, but as you note there are plenty that only look to the short term or don't have enough of an understanding to right the ship in terms of details.
However I do agree that the problem is always at the top first. It's just the solutions are rarely if ever easy.
 
Intel is heading into serious trouble if they don't right their ship soon.
 
Intel is heading into serious trouble if they don't right their ship soon.
Not really. It will take a long time of serious decline before there are massive issues.
Almost all of OEM sales and server sales are dominated by Intel. If Intel suddenly ceased to exist then 'the world' would be in serious trouble as quite frankly AMD and other niche manufacturers wouldn't be able to supply the void that Intel would leave behind.
If AMD sells every chip they make or it's possible for them to make, Intel still would have a lion's share of the market.

However, piece by piece the money will start to go other directions. Fujitsu released one of the most powerful super computers all using ARM. Apple is moving to ARM (they're relatively small, but roughly 10-15% of the consumer space). And AMD is continuing to show their server and workstation chips are worthy of real consideration.
Much like IBM, Blackberry, Xerox, etc, it will still take likely 10 years for a full decline like that to happen. Intel should be concerned. They should fix their problems now. Otherwise they have nothing left other than a long slow death. But make no mistake: we're just at the beginning of said long slow death.
 
At this point one has to wonder if there's another reason.

Generally I think the most likely reason that intel is in this has to do with the Core architecture. I have a feeling that most of those gains that they experienced came from:

1) L1/L2 cache speed
2) Lax security measures on their cache, memory controller, and branch predictors

So fast forward today once Intel started fixing the security problems they had only one thing left...cache speed, which for every shrink had been harder and harder to maintain.

Essentially there's no silver bullet. If Intel wants to beat AMD then it needs an entirely new architecture. This took AMD 5 years to do, with Jim Keller.

For Intel we could maybe cut that in half. But generally I think correcting past mistakes is going to increase that timeframe.

Intel needs a new core with process tech that will allow them to produce chips with low latency in cache while at the same time keeping power reasonable. If it relaxes the cache it's not going to beat AMD. So it's stuck until it can run everything at the high clocks and low latency and low power as it's used to doing.

It doesn't need 14nm+++ or 10 nm++++. It needs 7nm+++ now and today. Intel's process tech has always been no compromises. So what happens if you need to make compromises and your core is slower than before because of security updates? What do you do?

Well you need more performance that has been left on the table. But what if you waited to redo your core after AMD, IBM, and even Apple. Then what do you do? The answer isn't so simple and I have a feeling this is where Intel is at.
 
Last edited:
Not really. It will take a long time of serious decline before there are massive issues.
Almost all of OEM sales and server sales are dominated by Intel. If Intel suddenly ceased to exist then 'the world' would be in serious trouble as quite frankly AMD and other niche manufacturers wouldn't be able to supply the void that Intel would leave behind.
If AMD sells every chip they make or it's possible for them to make, Intel still would have a lion's share of the market.

However, piece by piece the money will start to go other directions. Fujitsu released one of the most powerful super computers all using ARM. Apple is moving to ARM (they're relatively small, but roughly 10-15% of the consumer space). And AMD is continuing to show their server and workstation chips are worthy of real consideration.
Much like IBM, Blackberry, Xerox, etc, it will still take likely 10 years for a full decline like that to happen. Intel should be concerned. They should fix their problems now. Otherwise they have nothing left other than a long slow death. But make no mistake: we're just at the beginning of said long slow death.


Aren't we already about 2 years in?
 
At this point one has to wonder if there's another reason.

Generally I think the most likely reason that intel is in this has to do with the Core architecture. I have a feeling that most of those gains that they experienced came from:

1) L1/L2 cache speed
2) Lax security measures on their cache, memory controller, and branch predictors

So fast forward today once Intel started fixing the security problems they had only one thing left...cache speed, which for every shrink had been harder and harder to maintain.

Essentially there's no silver bullet. If Intel wants to beat AMD then it needs an entirely new architecture. This took AMD 5 years to do, with Jim Keller.

For Intel we could maybe cut that in half. But generally I think correcting past mistakes is going to increase that timeframe.

Intel needs a new core with process tech that will allow them to produce chips with low latency in cache while at the same time keeping power reasonable. If it relaxes the cache it's not going to beat AMD. So it's stuck until it can run everything at the high clocks and low latency and low power as it's used to doing.

It doesn't need 14nm+++ or 10 nm++++. It needs 7nm+++ now and today. Intel's process tech has always been no compromises. So what happens if you need to make compromises and your core is slower than before because of security updates? What do you do?

Well you need more performance that has been left on the table. But what if you waited to redo your core after AMD, IBM, and even Apple. Then what do you do? The answer isn't so simple and I have a feeling this is where Intel is at.
Isn't ice lake taking over from Skylake?
 
Isn't ice lake taking over from Skylake?

Only for some mobile (now) and server (soon) parts. It's not going to be a mainstream desktop part ever. Their first 10nm desktop part: Alderlake isn't scheduled to be out until mid-next 2021. Although Ryan Smith is speculating that with Intel ramping 10nm production more due to 7nm slipping that it might be moved forward and launch before the end of the year, I'm rather skeptical.

https://www.anandtech.com/show/1592...alder-lake-in-h2-21-ice-lakesp-late-this-year
 
Ok, seems odd to have such a discrepancy for a longer period of time like this. I mean, AMD has a delay for their APU being behind their current microarchitecture, but they don't run a different architecture on mobile vs. desktop for multiple years (at least not that I can recall). I agree, they just released 10 series desktop parts and they seem to be selling, why would they try to rush a release by the end of year unless they are very concerned about zen3 and they don't want to be seen as falling further behind? If it's a rushed release it'll most likely be samples to reviewers and almost zero stock I'm sure just to try to tamp down excitement for zen3.
 
Something doesn't seem right. Maybe they've hit a wall with their current architecture? Maybe it's management?

It's weird seeing AMD catch up so fast and now might leap frog them?
 
Intel's 7nm is 'broken' while AMD's 7nm is thriving?...Intel doesn't seem like the same company that was dominating the CPU market for so long...Zen 3 is going to be the final foot stomp...maybe in 5 years Intel can recover but it's not going to be anytime soon...pick up an AMD CPU and don't look back
 
Something doesn't seem right. Maybe they've hit a wall with their current architecture? Maybe it's management?
It's the bets they made on process technology.
It's weird seeing AMD catch up so fast and now might leap frog them?
It shouldn't seem weird since it's TSMC that may 'leap frog' Intel. AMD has nothing without TSMC.
 
Intel's 7nm is 'broken' while AMD's 7nm is thriving?
Intel's 10nm is about the same feature density as TSMC's 7nm. Intel's 7nm is going to be about the same feature size as TSMC's 5nm.

TSMC exclusively manufactures outside designs, Intel exclusively manufactures their own designs. TSMC has a need to market their process technology to prospective customers, Intel does not.
 
Perhaps they tried and got the .... "sorry apple has our fab space reserved till 2023".

Didn't they suggest that they were going to consider using other Fabs while they worked out their 7nm difficulties which at this point would be either Samsung or TSMC
 
and Intel is still using 14nm (10nm is still 12+ months away) while AMD has been using 7nm for over a year
True, but the main point is that there's a bigger technology jump going on. TSMC going to their "7nm" process involved successfully making that jump, while Intel has struggled to make that jump going to their 10nm process.

AMD is simply lucky that TSMC chose correctly, and that's not something that AMD had any hand in. Similarly, Intel has stacks of processor architectures waiting for their manufacturing processes to catch up.
 
Back
Top