Discussion Intel current and future Lakes & Rapids thread

Page 277 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

inf64

Diamond Member
Mar 11, 2011
3,686
3,958
136
Intel's (and NVidia's) problem since 2017 is that they are fighting a much tougher and leaner enemy - it resembles a guerilla warfare to some extent. AMD has no more fabs to fund RND for, they have extremely talented overlapping engineering teams and they are racking in lot more $$$ than in the last 10 years - they can now redirect that money into uarchitecture design and talent hunting. They are also deep into consoles and probably locked next few gens with Sony and Microsoft (meaning it all gets optimized for Zen and RDNA derivatives). Intel and Nvidia (to lesser extent) have to execute flawlessly in the next 3-4 years to combat this new revival from AMD. Good luck to them!
 

A///

Diamond Member
Feb 24, 2017
4,352
3,154
136
Intel's (and NVidia's) problem since 2017 is that they are fighting a much tougher and leaner enemy - it resembles a guerilla warfare to some extent. AMD has no more fabs to fund RND for, they have extremely talented overlapping engineering teams and they are racking in lot more $$$ than in the last 10 years - they can now redirect that money into uarchitecture design and talent hunting. They are also deep into consoles and probably locked next few gens with Sony and Microsoft (meaning it all gets optimized for Zen and RDNA derivatives). Intel and Nvidia (to lesser extent) have to execute flawlessly in the next 3-4 years to combat this new revival from AMD. Good luck to them!

I'm not a gamer. Can you explain this to me? I know about AMD's Jaguar for the current active generation. Wouldn't that imply all console games that were also available on PC would have gotten that treatment? Or are you saying that because it is a direct copy of RDNA2 that it will be easier to optimize for AMD? I find it hard to believe.
 

SAAA

Senior member
May 14, 2014
541
126
116
I see everyone dubious on Rocket's boost speed, 5.5 GHz does sound extreme, but suppose the improvements they made into 10nm (Tiger lake's superFinfet) slipped up into 14nm whatever iteration it will be using?
They had to redesign most of the thing anyway by back-porting, I hear it's a whole "Cove" arch in itself, then it would make sense to have another 14nm change that allows even greater frequency.
 

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136
Good job calling out AMD in an Intel thread. Never mind that even the 1800x wasn't "factory overclocked" any more than Kaveri, Godavari, or desktop Carrizo (x4 945). Those chips had very little upper headroom due to process and/or cache latency (that was an issue for Carrizo, apparently). If you want to point to a "factory overclocked" CPU from AMD, you really have to go back to the FX9590 and 9370.

In any case, I don't think we'll see clocks from Rocket Lake-S that are as high as Comet Lake-S.
Not that anyone ever would hijack an Intel thread for pro AMD posts, but I never hear you call that out.
 
  • Like
Reactions: mikk

mikk

Diamond Member
May 15, 2012
4,112
2,108
136
I've been on Discord with Bepo for a long time. His predictions are far older than April this year.

1. You have to prove it
2. The predictions from him are unrelated
3. We are not talking about predictions, everyone could predict something, we are talking about a real source
4. If he or others predicted it right well done, but that's it.
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,575
146
1. You have to prove it
2. The predictions from him are unrelated
3. We are not talking about predictions, everyone could predict something, we are talking about a real source
4. If he or others predicted it right well done, but that's it.
I have to prove what? Look at MebiuW's profile on Weibo, you'll probably find an excel spreadsheet containing V/f curve points at some point. Look for one with a replies that say something akin to the following:

1600017080088.png

Then feel free to ask Bepo about that table yourself. Just ask him over Twitter. He'll tell you exactly what that table is. Do a little bit of research and you won't have to believe me but rather those directly involved.
 

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
Maybe. At that point it devolves into an argument of what chips are "factory overclocked". Regarding Rocket Lake-S, I expect power usage in line with the 10900K but with lower clocks (well, certainly not anything above 5 GHz). Does that make it a factory-overclocked chip?

To me, a chip is factory overclocked if a TDP is set and the chip willingly blows through that TDP without changes by the user. Intel chips currently do that. AMD chips do not.

You guys are nuts if you think RKL will hit 5.5 Ghz. More likely is that frequencies will be the same as Comet Lake, but with a focus on IPC.
 
  • Like
Reactions: Tlh97

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
I have to prove what? Look at MebiuW's profile on Weibo, you'll probably find an excel spreadsheet containing V/f curve points at some point. Look for one with a replies that say something akin to the following:

View attachment 29797

Then feel free to ask Bepo about that table yourself. Just ask him over Twitter. He'll tell you exactly what that table is. Do a little bit of research and you won't have to believe me but rather those directly involved.

I call BS on that statement. Tiger Lake boosts up to 4.8 Ghz. If what he claims is true the laptops it is going in would melt.
 
  • Like
Reactions: mikk

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,575
146
I call BS on that statement. Tiger Lake boosts up to 4.8 Ghz. If what he claims is true the laptops it is going in would melt.
This is MT CB15 iirc. Not single core.

And this is perfectly in line with the analysis done by the Lenovo product manager iirc.

EDIT (note that the 1165G7 caps out at 4.1GHz, whereas the 1185G7 will boost to 4.3GHz, which is where the real comparison lies. Regardless, it's within expectations.):

1600024315730.png
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I have to prove what? Look at MebiuW's profile on Weibo, you'll probably find an excel spreadsheet containing V/f curve points at some point. Look for one with a replies that say something akin to the following:

View attachment 29797

According to that statement, it makes sense why earlier leaks had Tigerlake at 4.3GHz. They might have been planning at 4.3GHz, but either felt they needed more, or was necessitated because it was going to be delayed.

I think 4.3GHz with launch in June/July would have been much better. The extra 10% doesn't make a whole lot of difference and you still would have got 10nm SF V/F advantages.
 
  • Like
Reactions: uzzi38 and Tlh97

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
To me, a chip is factory overclocked if a TDP is set and the chip willingly blows through that TDP without changes by the user. Intel chips currently do that. AMD chips do not.

You guys are nuts if you think RKL will hit 5.5 Ghz. More likely is that frequencies will be the same as Comet Lake, but with a focus on IPC.
I'd agree with this. Intel would have had to lengthen the execution pipelines to get higher clocks, and then latency and stalls would have a bad impact on performance per clock. Then, why up the IPC to begin with?
 
  • Like
Reactions: Tlh97

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I'd agree with this. Intel would have had to lengthen the execution pipelines to get higher clocks, and then latency and stalls would have a bad impact on performance per clock. Then, why up the IPC to begin with?

They are backporting, so it cannot be a dramatically different core.

If there are any imaginary frequency increases, it'll be because 14nm is 5 years old. And it'll use so much power to do it.

Increasing pipeline stages also come with core complexity increases which increase area and power used.
 

inf64

Diamond Member
Mar 11, 2011
3,686
3,958
136
I'm not a gamer. Can you explain this to me? I know about AMD's Jaguar for the current active generation. Wouldn't that imply all console games that were also available on PC would have gotten that treatment? Or are you saying that because it is a direct copy of RDNA2 that it will be easier to optimize for AMD? I find it hard to believe.

We are going off topic here but next gen consoles, Xbox X and PS5, are Zen(2) and RDNA2 based. All the games developed for those two next gen consoles will be Zen2/RDNA2 tuned. How that will reflect on PC is anyone's guess, but logical thing to assume is that AMD is going to have an advantage.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
We are going off topic here but next gen consoles, Xbox X and PS5, are Zen(2) and RDNA2 based. All the games developed for those two next gen consoles will be Zen2/RDNA2 tuned. How that will reflect on PC is anyone's guess, but logical thing to assume is that AMD is going to have an advantage.

People said the same thing about the current gen consoles, didn't materialize. CPU wise it may not matter because of Intel's process problems.
 
  • Like
Reactions: CHADBOGA

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
Not that anyone ever would hijack an Intel thread for pro AMD posts, but I never hear you call that out.

So you agree that you were taking the thread off-topic then? Glad we have an understanding.

To me, a chip is factory overclocked if a TDP is set and the chip willingly blows through that TDP without changes by the user. Intel chips currently do that. AMD chips do not.

TDP is kind of nebulous though, so it's hard to use that to determine whether or not a chip is "factory overclocked". Power specs are all complicated nowadays. For example, AMD CPUs with a TDP of 105W actually use 142-145W pretty consistently. Does it mean they're "factory overclocked" though? Not necessarily. And I'm not absolutely sure you can say the 9900k or 10900k are "factory overclocked" because they will back off their high power consumption after a certain period of time (the tau period). Chips like the FX9590 were "factory overclocked" because they were specifically binned for high leakage, high clocks, and constantly ran at clockspeeds well outside of the rational points of their voltage/clockspeed curve. And the FX9590 was a high-clocked carbon-copy of the FX8350. Same die from the same wafers.

The name of the game nowadays is to try to figure out how to opportunistically push CPUs outside of their optimal clockspeed/voltage range when the cooling solution can handle it. Overclockers kept finding OC headroom on chips that were conservatively-clocked to stay within certain power/efficiency specs (look at the 2600x, for example). I don't think we'll ever see "factory overclock" anything anymore. It's more like, for top-bin chips anyway, neither AMD nor Intel want to sell you any headroom. And they're loathe to do so on lower-end parts.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
Just going to remind everyone that "overclocking headroom" is in reality a gap of speed testing coverage. It is an engineering failure being marketed as a feature. That is why Intel chips often have a lot of headroom.

Well, there's headroom and then there's headroom. OC classics like the Celeron 300a, various Core2Duo/C2Q chips, Westmere Xeons, the 2600x, etc. then yes I think those might fall into that category. The 10900k probably has some OC headroom, but it punishes you with insane thermals to the point that you require massive cooling solutions. I doubt the engineers failed when they realized that releasing a 300W+ 10c chip would be a bad idea.
 
  • Like
Reactions: Tlh97 and scannall

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
They are backporting, so it cannot be a dramatically different core.

If there are any imaginary frequency increases, it'll be because 14nm is 5 years old. And it'll use so much power to do it.

Increasing pipeline stages also come with core complexity increases which increase area and power used.
I don't recall adding pipeline stages to be a major issue complexity wise (especially with modern tools and sims). But you have a point; Intel can just keep upping the current to maintain a halo CPU on RKL. They got a lot less backlash from the tech press than I thought they would (like, 'this uses too much power', but 'if you're a gamer, go with a 10900k'). The irony was pretty thick though.
 

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136

6-8-core 35-45W Tiger Lake for next year, though the GPU is slower.

EDIT: maybe we will see some SFF PCs with these chips in them.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
6-8-core 35-45W Tiger Lake for next year, though the GPU is slower.

Just to clarify, 35W TigerLake-H is only 4c according to the article. And if the 45W chip is supposed to compete with Cezanne, then . . . what happened to Alder Lake-P?
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
And if the 45W chip is supposed to compete with Cezanne, then . . . what happened to Alder Lake-P?
Isn't ADL more of a H2 2021 - H1 2022 kinda' product? I was under the impression the planning soup contained RKL + TGL for H1 2021.