Discussion Intel current and future Lakes & Rapids thread

Page 286 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

piokos

Senior member
Nov 2, 2018
554
206
86
AMD has a good line-up of laptops APU's that sells very well today and will continue to sell well since Renoir has gotten such great reputation. The last thing they should do is to mess this up by sudden changes that causes problems for the OEM's. This is not a "DYI-market competition".
AMD currently has a much bigger lead in desktops. Intel doesn't focus on this market, so AMD could keep selling Zen2 for a few months (maybe a year) with absolutely no impact on market share.

But of course DIY crowd would implode from not being able to spend another $1000 on a yearly benchmark result boost, so this is mostly about just taking care of the core client group. ;)

Laptops is another story. As of 2021, this will once again be an even competition. And Intel has a lot of other perks that will secure their market dominance.
Success among OEM's requires stability in the roadmap (among other things).
No, it doesn't. What it needs is close cooperation with OEMs. This is where AMD has failed for years.
Intel launches a new chip generation every year and most OEMs announce their implementations pretty much on the same day.
Isn't the question mark here, to some degree, Intel's ability to delivery TGL in volume? Can they even make enough chips so that we see this in every Thinkpad, Inspiron, etc. that it should be in?
Probably not. And there's no reason why that would ever happen. I mean: setting up fabs is a huge cost - they have to work for years. Intel will always offer a mix of nodes in their products.

Even Intel's 14nm - despite being around for many years - wasn't unified. They were making chips based on different variants of the node. Even 22nm was used from time to time.
Hunting for Ice Lake vs Comet Lake based laptops was rather infuriating based on their naming convention and it really soured me on it when I was helping folks find their college bound kids laptops this spring/summer.
That's just overthinking the problem. Most people buy a laptop based on features, performance, size, battery life and weight (more or less in that order). They don't care what exact parts are inside. And they shouldn't. OEM takes care of that.
There's hardly any difference between owning a Comet Lake and Ice Lake-based laptop if they have the same basic properties.

It's just that you, as an enthusiast, feel the need to have the latest stuff. ;)
 
  • Like
Reactions: Pilum

blckgrffn

Diamond Member
May 1, 2003
9,123
3,058
136
www.teamjuchems.com
Probably not. And there's no reason why that would ever happen. I mean: setting up fabs is a huge cost - they have to work for years. Intel will always offer a mix of nodes in their products.

Even Intel's 14nm - despite being around for many years - wasn't unified. They were making chips based on different variants of the node. Even 22nm was used from time to time.

That's just overthinking the problem. Most people buy a laptop based on features, performance, size, battery life and weight (more or less in that order). They don't care what exact parts are inside. And they shouldn't. OEM takes care of that.
There's hardly any difference between owning a Comet Lake and Ice Lake-based laptop if they have the same basic properties.

It's just that you, as an enthusiast, feel the need to have the latest stuff. ;)

I am not sure what variations 14 and 22nm matters to my question. Skylake was all a given process. As were others. And when we had 7th & 8th & 9th gen branding we knew what we were getting. The question was is if TGL is going be a clear launch because they can make enough for all the OEMs to get what they need or if it will be another co-mingled generation because their 10 capacity (whether the shortage is based on investment or yield ¯\_(ツ)_/¯) is insufficient.

As to just not caring what CPU is in a ~$1,000+ computer...

Well, that is swallowing the Intel line of "performance doesn't matter, the experience matters" line hard. There are tangible performance improvements in Ice Lake vs Comet Lake (it's a Tick+Tock, after all) especially in graphics. People don't come to me when picking hardware for general guidance but when they want help deciphering the platform configurations to get them the best setup for their money. Like determining if the SSD is NVME or SATA, memory is in dual channel mode, etc. Welcome to Anandtech, where details typically matter. Intel purposefully obfuscating their 10th gen offerings was a garbage move that hindered customers (including me!) ability to make an informed choice. Hopefully "11th Gen" or whatever the new branding is more clear cut.

I mean, that comment blows my mind. That's like saying whether someone bought a mobile i3 based on Nehalem vs one on Ivy Bridge if they were both available at the same time would be better made on which model had a back lit keyboard.
 
  • Like
Reactions: Tlh97 and Gideon

piokos

Senior member
Nov 2, 2018
554
206
86
I am not sure what variations 14 and 22nm matters to my question. Skylake was all a given process.
You asked if Intel can make enough Tiger Lake to put it in all mainstream laptops (that's how I understood it).
The answer is: no. And it simply can't happen. Because:
1) a node/fab has to be used for a particular time to turn a profit (I don't know how long... 5 years maybe?) - it can't be replaced earlier
2) they can't shut down a production line and instantly launch a new one - it's a transition and it takes time.

Nodes will always overlap. If you're TSMC, you can offer older fabs to consumers who don't need the latest and greatest.
If you're Intel, you have to make a product yourself. This is the main reason why Intel makes SSDs and network equipment - not because these are such fantastically profitable niches, but because they have to make something. They paid for the fab anyway.

Skylake (2015-) is made on 14nm, but in different generations ("+"). And AFAIK Intel doesn't really "port' designs to newest nodes. They usually make it the same way throughout its lifetime. This is different with AMD+TSMC, when you see new CPU batches getting better. Or old designs moved to new nodes completely ("Ryzen 5 1600 AF").

As of today, Intel still makes the original Skylake (6th gen) from 2015 and Apollo Lake from 2016. Probably the same way, on the same production line, they did on day 1. :)

What I meant was: 10nm volume will keep increasing, but realistically we should expect 14nm CPUs to be around until at least 2023. Next year 14nm will surely still be used for some of the top of the range desktop and laptop parts. Later probably just for low-end stuff.
The question was is if TGL is going be a clear launch because they can make enough for all the OEMs
As above: even if they could, they won't. 14nm has to keep making money. :)
As to just not caring what CPU is in a ~$1,000+ computer...
Yeah, I know you don't get this. I won't convince you.

But $1000 is like... MacBook Air / basic Dell XPS money. Nothing special.
I mean, that comment blows my mind. That's like saying whether someone bought a mobile i3 based on Nehalem vs one on Ivy Bridge if they were both available at the same time would be better made on which model had a back lit keyboard.
That's exactly how it works. Backlit keyboard, better screen, better choice of ports, better looks as well. That's the stuff you really interact with every day.

But only if key properties (performance, battery life, size etc) were at least close. This may not be the case for Nehalem vs Ivy Bridge, but for Comet Lake vs Ice Lake - absolutely. Putting graphics aside, they perform almost the same in daily use. And most Ice Lake laptops come with G1 and G4, not G7.

You know... I sometimes write this kind of non-nonsense stuff here and I see comments like this one. Total doubt and sarcasm. "Blows my mind" stuff.
No offense, but it's like some people here didn't have friends who don't game and don't run benchmarks... And I find that hard to believe.
 

blckgrffn

Diamond Member
May 1, 2003
9,123
3,058
136
www.teamjuchems.com
You asked if Intel can make enough Tiger Lake to put it in all mainstream laptops (that's how I understood it).
The answer is: no. And it simply can't happen. Because:

*Reasons*

That's a ton of rationale to explain away that 10nm continues to be a rolling disaster for Intel. Historically speaking, when Intel was functional, they would roll out new chips and quickly deprecate the old chips. That there was New-Old-Stock in the channel was to be expected, but they would put a shiny new generation badge on the new laptops and OEMs got new hype for their marketing and customers clearly knew they were getting something better than before.

The 8th-9th-10th generations for Intel are essentially making new SKUs available but not doing much with the underlying tech. More cores, less HT, then more HT again. Then they mixed the new laptop skus pretty hard into the 10th gen branding, which was a fairly unique marketing move.

And why? Because 10nm was such a mess they built out a fewer number of fabs and decided to focus on their 7nm process, which is now late too and further puts pressure on 10nm capacity and gives us a back port in the form of Rocket Lake on 14+++(++? ha). Which is simply an ongoing train wreck that I do understand.

The question I am asking is really if there will be enough Tiger Lake availability to matter, or will it be buried in a refresh full of other products named nearly the same thing, similar to Ice Lake. We're in a CPU thread on a fairly hardcore tech forum. There are people with their ear to the ground who might be able to say more interesting things like "Tiger Lake will ship at 2-3x the volume of Ice Lake for mobile. That means x-many units per month." That could be extrapolated into industry insight. I haven't read anything like that. I am not getting anything specific like that from recent responses.
Yeah, I know you don't get this. I won't convince you.

But $1000 is like... MacBook Air / basic Dell XPS money. Nothing special.

That's like an opinion, man. There are many deals to be had for those that know what they are looking for. Both of those purchases are heavy on brand, light on utility at that price point.

Also, I help a lot of people trying to stretch their dollars.

Earlier, you mentioned just trusting OEMs to provide good value and combinations of features at whatever price point. Clearly you know that there are SKUs being sold out there that are incredibly regrettable but eminently shiny. Or have no business being priced where they are but hey, why not charge $800 for a $500 item if you can get people to pay for it? Or maybe you can get last years hardware but since it's just a rebrand of this years hardware, it's a great value.

Which is the kind of information you expect your pc hardware nerd friend to know.

That's exactly how it works. Backlit keyboard, better screen, better choice of ports, better looks as well. That's the stuff you really interact with every day.

But only if key properties (performance, battery life, size etc) were at least close. This may not be the case for Nehalem vs Ivy Bridge, but for Comet Lake vs Ice Lake - absolutely. Putting graphics aside, they perform almost the same in daily use. And most Ice Lake laptops come with G1 and G4, not G7.

It sure is fun using ARK to see if the laptop is really a three year old model or a new one. My point isn't that the differences were great but that they exist. When you bought a nehalem or ivb CPU you knew this clearly at a glance by the model number. 10th generation whacked this, and we can differ on opinion but being on the latest platform is usually worth some premium. Intel must agree at least somewhat to co-brand 10th gen like they did. FWIW I ended up helping one of my friends by a P53s for his son this summer - with a Whisky Lake 8th Gen in it. But given it had great overall specs and big discount for being an outgoing model, the utility for the dollar was the most important.

You asserted that I only cared because I only want the latest and greatest. I'd disagree, I want to make sure I am not paying a premium for a product that is ancient, and if I can find the latest tech at same/lower prices I'd choose that instead.

You know... I sometimes write this kind of non-nonsense stuff here and I see comments like this one. Total doubt and sarcasm. "Blows my mind" stuff.
No offense, but it's like some people here didn't have friends who don't game and don't run benchmarks... And I find that hard to believe.

Haha, I do have friends like that. And sometimes their decisions do boggle my mind.

The thing is, they don't spend any time on Anandtech forums discussing the finer merits of e-die vs b-die ram or something else trivial to non-hobbysists because you're right, they don't care. Advocating that CPU generation/model doesn't matter in a CPU sub forum, along with advocating we trust OEMs to balance their products offerings fairly in terms of value is what is tripping my mental fuse.

Critically: They also don't make a point to ask me to personally shop tech for them. The people that do ask me want to make sure they are aren't buying (or buying another, based on past experiences) a SKU of Regret, as previously mentioned.
 
Last edited:

Dayman1225

Golden Member
Aug 14, 2017
1,152
974
146
Intel has launched additional Tigerlake SKUs today dedicated to the embedded/industiral market
  • i7 1185G7E
  • i5 1145G7E
  • i3 1115G4E
EpQzTDNFN6RXs9cNvYXpbE-970-80.png.webp

These feature reduced base/boost clocks vs the consumer versions and offer in-band EEC and -40c and 100c tempreature ratings.

Intel says they have >=90 partners committed to delivering products with these processors
 
  • Like
Reactions: Tlh97 and blckgrffn

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
According to Raichu little core Gracemont has a little better IPC than Skylake.

I think I called it. :p


It's the only way it made sense anyway. Skylake is actually not that special considering its 5 years old. Even back then it was a meh core.
 

cortexa99

Senior member
Jul 2, 2018
319
505
136
Intel has launched additional Tigerlake SKUs today dedicated to the embedded/industiral market
  • i7 1185G7E
  • i5 1145G7E
  • i3 1115G4E

These feature reduced base/boost clocks vs the consumer versions and offer in-band EEC and -40c and 100c tempreature ratings.

Intel says they have >=90 partners committed to delivering products with these processors

Finally, we get something that's other than 'skylake+14nm+++++' solution in the micro-PC or sub-PC market. Maybe it's not enough but it's a breakthrough still. (waiting skeleton)

But OTOH, a notebook sample which carry i7-1185G7 seems not performing well at power-consumption and temperature (by Linus Tech)

at 11:25 show the temperature is quite high and PL2 reach 64w......
 

piokos

Senior member
Nov 2, 2018
554
206
86
at 11:25 show the temperature is quite high and PL2 reach 64w......
Temperature is something for the OEM to take care of. But why would 64W of PL2 be a problem? Why are people so scared of boost power consumption? It's an advantage, not an issue.
 
  • Like
Reactions: TheGiant

coercitiv

Diamond Member
Jan 24, 2014
6,187
11,859
136
at 11:25 show the temperature is quite high and PL2 reach 64w......
We've been over this again and again, chip temps and short term power bursts are not a problem as long as they are intended behavior by the laptop manufacturer (they might increase BOM but let's worry about that when we see pricing). When evaluating laptops what matters is a combination of sustained performance, burst performance, skin temperature (what you feel on your lap, not internal temp), noise and battery life. The current TGL previews only show us half of the picture and we should simply wait for complete reviews.

As it stands right now we don't have final production firmware to compare versus the competition, and that means the final power/performance/thermal ratios are undetermined. We only have a rough estimate of how they influence each-other based on the pre-production Asus units that Intel provided to the press to preview.
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
Temperature is something for the OEM to take care of. But why would 64W of PL2 be a problem? Why are people so scared of boost power consumption? It's an advantage, not an issue.
exactly
instead of being happy of the possibility for a short time to use more power ...
I got used that Linus is the clickbait pisser and makes videos to get the audience but this is too much
amateur video
 
  • Like
Reactions: reb0rn

JoeRambo

Golden Member
Jun 13, 2013
1,814
2,105
136
It's the only way it made sense anyway. Skylake is actually not that special considering its 5 years old. Even back then it was a meh core.

Indeed even before 5 years it was money grab core, stingy on caches sizes and L2 was even nerfed to have 4-way instead of 8-way previous generations had. The only advances in instruction set AVX512 were locked in workstation variants. In retrospect it was horrible core for Intel to live with all those years. Basically made to milk customers with 4C 120mm2 design and then move on to 10nm.
How well did that work we saw ourselves :)

That's why AMD finally beating Skylake IPC in some applications is a feat only for them, customers already have such chips since 2015, only pricing and MT performance has advanced. Looking forward to ZEN3, that was hopefully not targeting to compete with 2015 designs anymore.
 
  • Like
Reactions: Tlh97 and Lodix

piokos

Senior member
Nov 2, 2018
554
206
86
exactly
instead of being happy of the possibility for a short time to use more power ...
I got used that Linus is the clickbait pisser and makes videos to get the audience but this is too much
amateur video
I could expect this to be partly a result of DIY mentality. You have to think about power consumption because of cooling, PSU, motherboard power delivery etc. I think it's understandable.
However, it's really surprising (and worrying) that people still don't understand boosting - years after it became so effective that it pretty much replaced overclocking in mainstream. For example: we keep seeing people who think a cooler has to match PL2. Bizarre.

During my time as a hardware geek and overclocker (15+ years ago), people seemed to understand computers a lot better. Or maybe I have this feeling because I was a part of more niche communities, mostly on Usenet (which may have been filtering out people who don't care that much...).
 

Antey

Member
Jul 4, 2019
105
153
116
During my time as a hardware geek and overclocker (15+ years ago), people seemed to understand computers a lot better.

well, you don't need a computer science degree to be computer entusiast. you can see lots of smartphone reviews with writers talking about IPC, frequencies, cores, arquitectures and such and with most probability they are just geeks writing for people that just like to talk and discuss about specs event tough they don't know much about it.

have you seen those guys in youtube comparing cpu speeds opening apps and seeing which one opens faster... linus is no thaaaaat different from that. he definitely is a clickbait pisser and makes videos to get the audience...

(and sorry for my english... had to say it!)
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,620
10,830
136
So, you interpret this quote as they can't and/or won't?

Yes. It currently doesn't exist on the open market, nor has it even been launched. We have no idea when that product will come to market. At this point, I take nothing for granted when it comes to Intel or AMD.
 
  • Like
Reactions: Tlh97 and tjf81

piokos

Senior member
Nov 2, 2018
554
206
86
well, you don't need a computer science degree to be computer entusiast.
No, but you should exhibit will to learn stuff and to understand the matter best you can.

I don't expect anyone to understand thermodynamics or semiconductors.
But I would expect everyone here to understand boosting. And to have a very basic, qualitative understanding of how heat accumulates and dissipates - especially when they spend so much time discussing it. It's secondary school level of understanding the world.
 
  • Like
Reactions: Tlh97 and A///

mikk

Diamond Member
May 15, 2012
4,133
2,136
136
The next generation Tiger Lake NUC has been detailed
 

TheELF

Diamond Member
Dec 22, 2012
3,973
730
126
The 8th-9th-10th generations for Intel are essentially making new SKUs available but not doing much with the underlying tech. More cores, less HT, then more HT again. Then they mixed the new laptop skus pretty hard into the 10th gen branding, which was a fairly unique marketing move.

And why? Because 10nm was such a mess they built out a fewer number of fabs and decided to focus on their 7nm process, which is now late too and further puts pressure on 10nm capacity and gives us a back port in the form of Rocket Lake on 14+++(++? ha). Which is simply an ongoing train wreck that I do understand.
Well back when intel was functional AMD was in such terrible shape that intel had to keep with 4/8 CPUs in mainstream or they would have wiped AMD out completely, so being unable to change the core number and HTT the only thing they could do was to release a new node every now and then to make people upgrade their systems.
Now that AMD is doing a little better intel can go crazy with cores and HTT and it doubled intel's net income for the last two years.


Der8auer took pictures of intel 14nm and tsmc 7nm and guess what, there is only a minimal difference in favor of tsmc not even worth all the ruckus.
Intel's 10nm might be a mess but it looks more like intel is holding back their better nodes until they have to release them.Intel is using less cores each core with less compute units with less cache and their wafers produce at least roughly the same amount of product as tsmc's 7nm, intel is rofling all the way to the bank here.

KYz5FnL.jpg
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
I could expect this to be partly a result of DIY mentality. You have to think about power consumption because of cooling, PSU, motherboard power delivery etc. I think it's understandable.
However, it's really surprising (and worrying) that people still don't understand boosting - years after it became so effective that it pretty much replaced overclocking in mainstream. For example: we keep seeing people who think a cooler has to match PL2. Bizarre.

During my time as a hardware geek and overclocker (15+ years ago), people seemed to understand computers a lot better. Or maybe I have this feeling because I was a part of more niche communities, mostly on Usenet (which may have been filtering out people who don't care that much...).
oh Usenet, thanks for reminder (2000-2005 times for me)
well the main gap I see is disconnection of the benchmark results from real uses
or translation- internet heroes don't actually do anything, so there is nothing to connect to
run benchmarks is the computers use
and then ...........another benchmark
the internet is brainwashed by so called throughput benchmarksin every segment, even if its far from that segment primary parameter
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
Well back when intel was functional AMD was in such terrible shape that intel had to keep with 4/8 CPUs in mainstream or they would have wiped AMD out completely, so being unable to change the core number and HTT the only thing they could do was to release a new node every now and then to make people upgrade their systems.
Now that AMD is doing a little better intel can go crazy with cores and HTT and it doubled intel's net income for the last two years.


Der8auer took pictures of intel 14nm and tsmc 7nm and guess what, there is only a minimal difference in favor of tsmc not even worth all the ruckus.
Intel's 10nm might be a mess but it looks more like intel is holding back their better nodes until they have to release them.Intel is using less cores each core with less compute units with less cache and their wafers produce at least roughly the same amount of product as tsmc's 7nm, intel is rofling all the way to the bank here.

KYz5FnL.jpg
yeah, noticed that article
there is only one wrinkle though- the power difference
what exactly does that article tell us ? how is that info usable
 
  • Like
Reactions: Tlh97

blckgrffn

Diamond Member
May 1, 2003
9,123
3,058
136
www.teamjuchems.com
Now that AMD is doing a little better intel can go crazy with cores and HTT and it doubled intel's net income for the last two years.

Hey man, let's go!

Honestly I am excited about TGL and Xe, I would just appreciate it if it was clearly delineated from other CPUs and widely available soon. If it could become near ubiquitous between TGL 11th Gen vs AMD Renoir 4xxx series as a baseline I think all consumers win. Perhaps six months from now? Looking at next summer & fall purchases...
 
  • Like
Reactions: Tlh97

DrMrLordX

Lifer
Apr 27, 2000
21,620
10,830
136
Well back when intel was functional AMD was in such terrible shape that intel had to keep with 4/8 CPUs in mainstream or they would have wiped AMD out completely

Intel did that to keep ASPs high without increasing die size on consumer parts.

Now that AMD is doing a little better intel can go crazy with cores and HTT and it doubled intel's net income for the last two years.

That's not how that works.

Intel's 10nm might be a mess but it looks more like intel is holding back their better nodes until they have to release them

That's definitely not how that works.

Intel is using less cores each core with less compute units with less cache and their wafers produce at least roughly the same amount of product as tsmc's 7nm, intel is rofling all the way to the bank here.

Intel makes bank from consumer lock-in. Their sales department is keeping them alive and well. Don't think that 14nm is without major problems. XCC and HCC dice are (or have been) eating up 14nm wafers, and Intel still can't move everything off 14nm yet. They aren't even close.

TSMC is winning on transistor density vs Intel 14nm. Go look at transistor count vs die area on any of Intel's 14nm CPUs vs. something like a Rome chiplet, and you will see what i mean.
 
Last edited:

eek2121

Platinum Member
Aug 2, 2005
2,930
4,026
136
Hey man, let's go!

Honestly I am excited about TGL and Xe, I would just appreciate it if it was clearly delineated from other CPUs and widely available soon. If it could become near ubiquitous between TGL 11th Gen vs AMD Renoir 4xxx series as a baseline I think all consumers win. Perhaps six months from now? Looking at next summer & fall purchases...
10SF is in the middle of a ramp. Due to the way product cycles work, we will see one more 14nm performance launch. After that Most of Intel’s performance products will be on 10nm. When Rocket Lake was being designed, 10nm was in bad shape.
 

ondma

Platinum Member
Mar 18, 2018
2,720
1,280
136
Intel did that to keep ASPs high without increasing die size on consumer parts.
Now that AMD is doing a little better intel can go crazy with cores and HTT and it doubled intel's net income for the last two years.[/quote]

That's not how that works.



That's definitely not how that works.



Intel makes bank from consumer lock-in. Their sales department is keeping them alive and well. Don't think that 14nm is without major problems. XCC and HCC dice are (or have been) eating up 14nm wafers, and Intel still can't move everything off 14nm yet. They aren't even close.

TSMC is winning on transistor density vs Intel 14nm. Go look at transistor count vs die area on any of Intel's 14nm CPUs vs. something like a Rome chiplet, and you will see what i mean.
[/QUOTE]
News flash, a new process is winning against a 5 year old one? Who would have ever guessed?