"company is announcing that their 7nm CPU plans have been pushed back by 6 months" new Intel cadence (evolution of tick-tock): new march, backporting, new process or backporting, surely new process
"So while 14nm has been something of an albatross for Intel, it’s not going to be a short-lived manufacturing node as Intel rushes to 22nm; instead it’s going to be a core part of their plans for the next two or three years."
ha ha ha ... over nine years for notel to get back to 22nm++++++ but $$$ in the meantime due to legacy dependency on COBOL
They were hoping for this to take them away from the 10nm woes, which are ongoing.
Now Intel is going to have to rely on 10nm for the next two years, possibly longer, as their leading edge node, despite, by their own admission, it not being very good. It seems the tweaks for this year are going to enable TGL to hit a turbo of 4.7GHz which is far better than the IceLake 4.1GHz (did anyone ever see that?), but not near 14nm. IPC to the rescue - except that goes for AMD as well. And yield - this is going to increase their costs, and we see their gross margin dropping. Intel can't raise prices because of AMD being competitive finally. It's going to be a rough year or two, but they have a lot of momentum/inertia to help.
which 10nm woes? About to release what will be an ultra high volume Ice Lake SP Xeon, Tiger Lake and the discrete graphics card - all on 10nm+. 10nm has been in the market since September of 2019. 10nm issues are long gone.
The 7nm delay is troublesome - but their Xe HPC will be produced at 7nm regardless of the yields - it's a halo product. Moving to full EUV (and not just a few layers like TSMC) and integrating cobalt into the M0/M1 layers is a heavy lift. If Intel is running into issues with full stack EUV you can bet that TSMC will as well. Samsung has a 2 year head start working with EUV and they are having issues. Was dubious about having the 7nm shrinks of Alder Lake (Meteor Lake) and Sapphire Rapids (Granite Rapids) - so relatively quickly - so 10nm will be a longer node than expected.
IPC on the Tiger Lake are well past the IPC lift from AMD - when a 4C Tiger Lake can only lose by 6% in windows vs a 8C 4800U - that will be an issue for AMD - also gets pwned pretty much completely in the iGPU dept.
Looks like that 4.1 for Ice Lake and 4.7 for Tiger Lake to be close to the max that AMD's desktop CPUs can hit - so parity there, but 14nm has been frequency optimized for several refreshes - getting close to 14nm frequency with massively improved IPC mean higher performance.
All of this, while another record quarter driven by data center and laptop - is a win for Intel - and still waiting for this AMD competition - and a sky high PE (105x) does not create momentum or intertia - it's an optimism from Wall Street - but at some point AMD has to deliver - which is hasn't yet. Intel posts record quarter after record quarter and Wall Street yawns - because they are so detached from actual results and are instead playing casino. Competition requires Sales.
LOL, Intel is funny, they want to gain money like mad max. When a process looks expensive, they truncate it and say some idiocy like "a defect in their 7nm process, which the company has since root caused and is correcting". All this is hard to believe, both 10nm and 7nm processes are fixed in stone from ages. The real suspect is: the ASML equipment is reliable enough?? The process stack have an economic sense?? The real question is : why Samsung is having so much troubles on 7nm EUV and 5nm EUV ??
TSMC looks out of these waters but.....hey !! the insane capex !! HUGE for a not so large company like TSMC. Ummmm something is wrong, actually TSMC looks like a no profit company at the service of customers but without desire of real money gains.
That's funny, AMD was supposed to be absolutely amazing in server space with all those multi-core EPYC/whatever shits but hey, Intel is having another record quarter I wonder why? Oh, must be because nobody serious actually uses AMD in servers (or on desktop... or anywhere... sounds familiar?)...
Nobody uses AMD anywhere... but AMD are having their most profitable quarters since 2001... Okay, timcarp! You're the genius here!
It's scale, stupid. Everybody and their dog knows AMD won't be displacing Intel out of their core markets any time soon - even if there was no drag from market cycles, they don't have the ability to output anywhere near the required volume of products!
The point is that they finally have superior tech *to begin an assault on that market*, and bow howdy is it superior. You can tell by how salty the shills get.
Well compared to the Dumpter fires before Zen launch they would have to be doing better than that - Still waiting on that AMD Competition - if it were competitive, then their revenue would be as flat as it has been since Zen 1 Refresh. Still making no money - certainly not enough to fund a R&D department and pay Lisa Su's exorbitant salary.
I do agree that AMD won't be displacing Intel out it's core markets - Desktop, Laptop, Server.
Where is this magical superior tech?
Also, the term is "boy howdy" No shills are as salty as you and your little "friend"...
If you don't see it, you've not been paying attention. AMD *has* been gaining traction in servers. I'm using three VMs on it myself, up from zero this time last year. Yes, there's still a lot of Intel gear in the market, and it's not going away any time soon. But they've made real in-roads, at Intel's expense, in part because Intel just couldn't deliver and clouds were willing to take a chance on it.
Similarly for laptops - even if they're there in part to extract concessions out of Intel, they can be ordered and are finally competitive, if still not quite there with single-threaded perf. (Then again, with all the mitigations...)
The same is happening here. Our latest tech refresh is replacing our current tired servers with new hardware and we are purchasing PowerEdge servers that all run on Epyc CPUs. Intel does not offer a competitive product at the prices and performance curve our budget folks will accept so at least in our little datacenter there will be a few thousand AMD cores in place by year end.
And they claimed to have "fixed" those issues a whole year before Ice Lake released. They "fixed" them so good that they had to back-port Sunny Cove to 14nm just to be able to get a new architecture out on desktop for the first time in 5 years.
Rocket Lake S has been in the slides for ages - was unnamed at the time, was listed as the last 14nm desktop with at that time unnamed architecture - We didn't know Sunny Cove and certainly not Willow Cove. So, been planned for some time, and will use Willow Cove with a modified cache structure, not Sunny Cove.
So far AMD still hasn't been able to put an appreciable dent in Intel's desktop lineup...
This bit made me chuckle: "So while 10nm has been something of an albatross for Intel, it’s not going to be a short-lived manufacturing node as Intel rushes to 7nm; instead it’s going to be a core part of their plans for the next two or three years."
We went from ~2 years of repeating "10nm is gonna be fine, honest" to a year of "Cannon Lake totally launched but please don't ask us about it and 10nm is totally fine" to a year of "ackshually we're leveraging 14nm+++++ because of how aewsoem it is" before finally arriving at something that almost looked like honesty with "okay so 10nm isn't all that and a bag of chips, but it's a short node, onwards to 7nm!"
Now they want us to buy "7nm is fine but also 10nm is gonna be a long node now"? Good play, guys, good play.
At least it's not doing any damage to their financials. I'll be interested to see what happens to the market as a whole when the coronavirus-inspired surge in demand for cloud compute power and laptops slackens, though.
I don't really get the Intel pessimism. The surge of income right as Intel finally had a moment of weakness should bandage over their problems. Decoupled CPU modules with higher-size IO die -- the chief reason AMD isn't having yield problems on core counts that high -- should push Intel into dramatically higher yields. Give them a few years to bring that to market and they'll be back on top.
"Give them a few years to bring that to market and they'll be back on top."
If that's the case then it's clear the market needs a third major player. Intel sold chips riddled with security flaws and couldn't execute past 14nm. There should be consequences for these failures but, instead, Intel rakes in the cash.
It must be nice to fail upwardly. I suppose that is what happens in duopoly, the thing that dominates tech.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
23 Comments
Back to Article
TristanSDX - Thursday, July 23, 2020 - link
"company is announcing that their 7nm CPU plans have been pushed back by 6 months"new Intel cadence (evolution of tick-tock):
new march, backporting, new process or backporting, surely new process
Everett F Sargent - Thursday, July 23, 2020 - link
typo ... should be ..."So while 14nm has been something of an albatross for Intel, it’s not going to be a short-lived manufacturing node as Intel rushes to 22nm; instead it’s going to be a core part of their plans for the next two or three years."
ha ha ha ... over nine years for notel to get back to 22nm++++++ but $$$ in the meantime due to legacy dependency on COBOL
psychobriggsy - Thursday, July 23, 2020 - link
The 7nm delay is a disaster for Intel.They were hoping for this to take them away from the 10nm woes, which are ongoing.
Now Intel is going to have to rely on 10nm for the next two years, possibly longer, as their leading edge node, despite, by their own admission, it not being very good. It seems the tweaks for this year are going to enable TGL to hit a turbo of 4.7GHz which is far better than the IceLake 4.1GHz (did anyone ever see that?), but not near 14nm. IPC to the rescue - except that goes for AMD as well. And yield - this is going to increase their costs, and we see their gross margin dropping. Intel can't raise prices because of AMD being competitive finally. It's going to be a rough year or two, but they have a lot of momentum/inertia to help.
Deicidium369 - Thursday, July 23, 2020 - link
which 10nm woes? About to release what will be an ultra high volume Ice Lake SP Xeon, Tiger Lake and the discrete graphics card - all on 10nm+. 10nm has been in the market since September of 2019. 10nm issues are long gone.The 7nm delay is troublesome - but their Xe HPC will be produced at 7nm regardless of the yields - it's a halo product. Moving to full EUV (and not just a few layers like TSMC) and integrating cobalt into the M0/M1 layers is a heavy lift. If Intel is running into issues with full stack EUV you can bet that TSMC will as well. Samsung has a 2 year head start working with EUV and they are having issues. Was dubious about having the 7nm shrinks of Alder Lake (Meteor Lake) and Sapphire Rapids (Granite Rapids) - so relatively quickly - so 10nm will be a longer node than expected.
IPC on the Tiger Lake are well past the IPC lift from AMD - when a 4C Tiger Lake can only lose by 6% in windows vs a 8C 4800U - that will be an issue for AMD - also gets pwned pretty much completely in the iGPU dept.
Looks like that 4.1 for Ice Lake and 4.7 for Tiger Lake to be close to the max that AMD's desktop CPUs can hit - so parity there, but 14nm has been frequency optimized for several refreshes - getting close to 14nm frequency with massively improved IPC mean higher performance.
All of this, while another record quarter driven by data center and laptop - is a win for Intel - and still waiting for this AMD competition - and a sky high PE (105x) does not create momentum or intertia - it's an optimism from Wall Street - but at some point AMD has to deliver - which is hasn't yet. Intel posts record quarter after record quarter and Wall Street yawns - because they are so detached from actual results and are instead playing casino. Competition requires Sales.
Korguz - Thursday, July 23, 2020 - link
oh i dont know, maybe these :https://www.tomshardware.com/news/intel-announces-...
https://www.tomshardware.com/news/intel-says-first...
intel is still doing ok huh? intel has no more issues with its 10nm or anything after ? i call BS on your part, Deicidium369
arashi - Sunday, July 26, 2020 - link
You have to forgive him, Intel marketing is in overdrive to try and keep their jobs.Teckk - Friday, July 24, 2020 - link
The 10nm which was delayed by 4 years.brantron - Thursday, July 23, 2020 - link
4.7 GHz is the same as the mobile and Xeon parts all of the new 10nm CPUs will replace.Average clock speeds are likely on the way up.
Gondalf - Thursday, July 23, 2020 - link
LOL, Intel is funny, they want to gain money like mad max. When a process looks expensive, they truncate it and say some idiocy like "a defect in their 7nm process, which the company has since root caused and is correcting".All this is hard to believe, both 10nm and 7nm processes are fixed in stone from ages. The real suspect is: the ASML equipment is reliable enough?? The process stack have an economic sense??
The real question is : why Samsung is having so much troubles on 7nm EUV and 5nm EUV ??
TSMC looks out of these waters but.....hey !! the insane capex !! HUGE for a not so large company like TSMC. Ummmm something is wrong, actually TSMC looks like a no profit company at the service of customers but without desire of real money gains.
Well time for Intel to be a TSMC customer too.
brucethemoose - Thursday, July 23, 2020 - link
TSMC has a bigger market cap than Intel. Not as much revenue, but they have a gross margin of ~50% too.Spunjji - Friday, July 24, 2020 - link
Classic Gondalf! All the words, none of the meaning.FreckledTrout - Thursday, July 23, 2020 - link
AMD has killed Intel's margins :)timecop1818 - Friday, July 24, 2020 - link
That's funny, AMD was supposed to be absolutely amazing in server space with all those multi-core EPYC/whatever shits but hey, Intel is having another record quarter I wonder why? Oh, must be because nobody serious actually uses AMD in servers (or on desktop... or anywhere... sounds familiar?)...Spunjji - Friday, July 24, 2020 - link
Nobody uses AMD anywhere... but AMD are having their most profitable quarters since 2001...Okay, timcarp! You're the genius here!
It's scale, stupid. Everybody and their dog knows AMD won't be displacing Intel out of their core markets any time soon - even if there was no drag from market cycles, they don't have the ability to output anywhere near the required volume of products!
The point is that they finally have superior tech *to begin an assault on that market*, and bow howdy is it superior. You can tell by how salty the shills get.
Deicidium369 - Friday, July 24, 2020 - link
Well compared to the Dumpter fires before Zen launch they would have to be doing better than that - Still waiting on that AMD Competition - if it were competitive, then their revenue would be as flat as it has been since Zen 1 Refresh. Still making no money - certainly not enough to fund a R&D department and pay Lisa Su's exorbitant salary.I do agree that AMD won't be displacing Intel out it's core markets - Desktop, Laptop, Server.
Where is this magical superior tech?
Also, the term is "boy howdy" No shills are as salty as you and your little "friend"...
GreenReaper - Friday, July 24, 2020 - link
If you don't see it, you've not been paying attention. AMD *has* been gaining traction in servers. I'm using three VMs on it myself, up from zero this time last year. Yes, there's still a lot of Intel gear in the market, and it's not going away any time soon. But they've made real in-roads, at Intel's expense, in part because Intel just couldn't deliver and clouds were willing to take a chance on it.Similarly for laptops - even if they're there in part to extract concessions out of Intel, they can be ordered and are finally competitive, if still not quite there with single-threaded perf. (Then again, with all the mitigations...)
PeachNCream - Sunday, July 26, 2020 - link
The same is happening here. Our latest tech refresh is replacing our current tired servers with new hardware and we are purchasing PowerEdge servers that all run on Epyc CPUs. Intel does not offer a competitive product at the prices and performance curve our budget folks will accept so at least in our little datacenter there will be a few thousand AMD cores in place by year end.brucethemoose - Thursday, July 23, 2020 - link
Well thats bad. Perhaps some of the process woes can be offset by gratuitous use of advanced packaging the time around.Of course, a small delay is exactly how the 10nm kerfuffle started...
Spunjji - Friday, July 24, 2020 - link
And they claimed to have "fixed" those issues a whole year before Ice Lake released. They "fixed" them so good that they had to back-port Sunny Cove to 14nm just to be able to get a new architecture out on desktop for the first time in 5 years.Deicidium369 - Friday, July 24, 2020 - link
Rocket Lake S has been in the slides for ages - was unnamed at the time, was listed as the last 14nm desktop with at that time unnamed architecture - We didn't know Sunny Cove and certainly not Willow Cove. So, been planned for some time, and will use Willow Cove with a modified cache structure, not Sunny Cove.So far AMD still hasn't been able to put an appreciable dent in Intel's desktop lineup...
Spunjji - Friday, July 24, 2020 - link
This bit made me chuckle:"So while 10nm has been something of an albatross for Intel, it’s not going to be a short-lived manufacturing node as Intel rushes to 7nm; instead it’s going to be a core part of their plans for the next two or three years."
We went from ~2 years of repeating "10nm is gonna be fine, honest" to a year of "Cannon Lake totally launched but please don't ask us about it and 10nm is totally fine" to a year of "ackshually we're leveraging 14nm+++++ because of how aewsoem it is" before finally arriving at something that almost looked like honesty with "okay so 10nm isn't all that and a bag of chips, but it's a short node, onwards to 7nm!"
Now they want us to buy "7nm is fine but also 10nm is gonna be a long node now"? Good play, guys, good play.
At least it's not doing any damage to their financials. I'll be interested to see what happens to the market as a whole when the coronavirus-inspired surge in demand for cloud compute power and laptops slackens, though.
lmcd - Sunday, July 26, 2020 - link
I don't really get the Intel pessimism. The surge of income right as Intel finally had a moment of weakness should bandage over their problems. Decoupled CPU modules with higher-size IO die -- the chief reason AMD isn't having yield problems on core counts that high -- should push Intel into dramatically higher yields. Give them a few years to bring that to market and they'll be back on top.Oxford Guy - Saturday, August 1, 2020 - link
"Give them a few years to bring that to market and they'll be back on top."If that's the case then it's clear the market needs a third major player. Intel sold chips riddled with security flaws and couldn't execute past 14nm. There should be consequences for these failures but, instead, Intel rakes in the cash.
It must be nice to fail upwardly. I suppose that is what happens in duopoly, the thing that dominates tech.