Intel Skylake-S CPUs to support new LGA 1151 socket

[Geekazoid]

it makes sense, but im not saying its the only explanation. its just the best one i can come up with, at this moment.

Well it may not be the only explanation, but it's good enough for me.

BTW, just thought I'd ask, how're ya going Luka?

[LukaP]

Well it may not be the only explanation, but it's good enough for me.

BTW, just thought I'd ask, how're ya going Luka?

fair enough.

 

oh im fine kinda busy, but otherwise normal. you

[OhhTee]

I will jjst wait for it to arrive and then I am gonna buy it.

I like the way you think.

[Geekazoid]

fair enough.

 

oh im fine kinda busy, but otherwise normal. you

Sweet! Yeah, kinda not busy , but good.

[bomerr]

i agree with him, but its been quite a known rumour for a while that intel are skipping broadwell. dont ask me why, but i dont have a problem with that.

 

im guessing they either know 10nm will be the last node before they have to switch to new materials/quantum processing, and they want it out fast so they have more time to focus on developing the future, or they are being pressured by big clients to deliver some new technology that skylake will have. (its not always the sales money that is the one you should look at)

 

Corporations want smooth steady growth, having a year of great sales followed by a year or two of bad sales KILLS stock price and gets management in deep trouble. If anything that would want to do the opposite. Stall as much as possible and string out your revenue while working on the next project. 

 

Apple does a great job of this with their iPhone XS upgrades. Likewise Intel has been doing this somewhere with their tick-tock schedule but even more so with the Haswell refresh where they reused the same architecture but clocked the chips higher.

 

Financially speaking doing more refreshes instead of full on upgrades makes the best sense. i mean look at that timeline, 6 years is all Intel has left of growth and profits. I'd be hella worried about that if I was upper management. 

[Geekazoid]

I will jjst wait for it to arrive and then I am gonna buy it.

You can be one of the early consumer testers.

[Recon97]

You can be one of the early consumer testers.

Maybe

[Recon97]

I like the way you think.

Thnx

[Geekazoid]

Maybe

It's better for you to waste your money than for everyone else to waste theirs. (just joking)

[deathjester]

Now I regret getting a Z97 mobo. -.-

 

Why? The MB you have can take 2 top end GPU's just fine.  Skylake will prob be like everything past Sandy Bridge. Lower top end OC on average, higher clock for clock that ends up very close and uses less power. It MIGHT get people to move off Sandy Bridge I5 K's, but I highly doubt it. Maybe the guys who got low clocking ones or the ones who haven't upgraded mb/cpu since 2011 and are like why not.

 

If you are doing a new build? Of course get it when it is out, just like you get Haswell now. As an enthusiast who knows how to overclock? You prob aren't missing a whole lot unless you planned to put 3-4 GPU's in your system. DDR4 prob won't be great or a good value for years yet.

 

I don't know if I would "wait" for Skylake unless I was on pre Sandy Bridge or a non K sandy Bridge. Especially when the 4770k was 199 bucks at Microcenter last Black Friday. People price matched it through Staples Online. Grab one of those if they do the same thing again this year, or if they do it with the 4790k, or if the I5 k reaches 150 again and call it a day. Low level API will make that CPU overkill for gaming as it is.

 

GPU jumps have been way bigger than CPU jumps the last few years. 

[Recon97]

It's better for you to waste your money than for everyone else to waste theirs. (just joking)

It depends.If its good,then only I am going to buy.

[Geekazoid]

It depends.If its good,then only I am going to buy.

True that!

[LukaP]

Corporations want smooth steady growth, having a year of great sales followed by a year or two of bad sales KILLS stock price and gets management in deep trouble. If anything that would want to do the opposite. Stall as much as possible and string out your revenue while working on the next project. 

 

Apple does a great job of this with their iPhone XS upgrades. Likewise Intel has been doing this somewhere with their tick-tock schedule but even more so with the Haswell refresh where they reused the same architecture but clocked the chips higher.

 

Financially speaking doing more refreshes instead of full on upgrades makes the best sense. i mean look at that timeline, 6 years is all Intel has left of growth and profits. I'd be hella worried about that if I was upper management. 

well if they get 10nm out, then they can milk it for ages, while developing graphene/quantum/whatever, so smooth is still there.

 

thats just pure and utter bull... as long as there is need for High intensity compute, intel has room to grow and profit. just look at them, optimising power and IPC at the same time. scaling FUCKING X86 into a <4W product. this is some next level stuff man. and i assure you, they are revealing as little of their aces as they can

[patrickjp93]

From a financial point of view it doesn't make any sense to skip broadwell. Intel can only do so many more die shrinks before they are done and skipping broadwell to go straight to skylake kills atleast a full years worth of profits. 

 

In all likelihood broadwell will come out next summer or fall and skylake will be pushed back until '16

Nope, I can tell you for certain Skylake is end of '15 Q2, S launches alongside Broadwell K, and there is no Broadwell S. That was the easiest way to avoid cannibalizing sales. Also, Intel still has 10, 7, 5, 3 nm in its roadmaps for silicon. IBM is also nearing 2nm for carbon nanotubes.

[bomerr]

well if they get 10nm out, then they can milk it for ages, while developing graphene/quantum/whatever, so smooth is still there.

 

thats just pure and utter bull... as long as there is need for High intensity compute, intel has room to grow and profit. just look at them, optimising power and IPC at the same time. scaling FUCKING X86 into a <4W product. this is some next level stuff man. and i assure you, they are revealing as little of their aces as they can

What would they do? Have a 10nm refresh in year 2 then again 3 and again in 4? People would complain there is little change and profits would shrink as people chose not to upgrade their prior CPUs. 

 

Haswell refresh was kinda nice because they simply overlocked the CPUs higher toward their maximum potentials but that was a one time deal. There won't and can't be 2 years of refreshes without serious effects on profit let alone 3 or 4. 

 

And no intel doesn't have some magical ace up their sleeves. Everyone in the semiconductor industry knows that before 2030 the systematic, year over year, improvements in CPUs will stop. 

[patrickjp93]

What would they do? Have a 10nm refresh in year 2 then again 3 and again in 4? People would complain there is little change and profits would shrink as people chose not to upgrade their prior CPUs. 

 

Haswell refresh was kinda nice because they simply overlocked the CPUs higher toward their maximum potentials but that was a one time deal. There won't and can't be 2 years of refreshes without serious effects on profit let alone 3 or 4. 

 

And no intel doesn't have some magical ace up their sleeves. Everyone in the semiconductor industry knows that before 2030 the systematic, year over year, improvements in CPUs will stop. 

You have to remember the amount of die space that is bought by the 10nm shrink from 14. 38%! Remember the die size of Westmere? Intel can milk a die size for years without shrinking merely by adding transistors.

[bomerr]

You have to remember the amount of die space that is bought by the 10nm shrink from 14. 38%! Remember the die size of Westmere? Intel can milk a die size for years without shrinking merely by adding transistors.

The issue is if you don't shrink the die then you can't add more resistors without changing the architecture. And optimizing the architecture for the specific die size is harder and harder to do each time you improve it. 

 

What'll end up happening, inevitably after 2025, is they will start increasing the component size to add in more transitions because it'll be the only way to increase power but obviously heat output, power consumption, and physical size will all suffer because of that. 

[LukaP]

What would they do? Have a 10nm refresh in year 2 then again 3 and again in 4? People would complain there is little change and profits would shrink as people chose not to upgrade their prior CPUs. 

 

Haswell refresh was kinda nice because they simply overlocked the CPUs higher toward their maximum potentials but that was a one time deal. There won't and can't be 2 years of refreshes without serious effects on profit let alone 3 or 4. 

 

And no intel doesn't have some magical ace up their sleeves. Everyone in the semiconductor industry knows that before 2030 the systematic improvements in CPUs will stop. 

skylake is 14nm. that is 2015 covered. skylake refresh is 14nm skylake, so h1 2016, then comes cannonlake on 10nm, to h1 2017, where cannonlake refresh comes in. then you have a new arch, still on 10nm. this is somewhere in q4 17. then in q2 18 you have that architecture refreshed. and in that amount of time, i can guarantee you, intel will have something exotic again. they always do

 

everyone knew that same thing when 100nm was to be passed. everyone said the leakage was too high. i assure you that intel have enough aces up their sleeves to get through anything mother nature throws at them at this time

[Techhog]

Is that 32 lanes on a skylake CPU that is not enthusiast?

 

OOOOOH GOD

That might mean that we'll finally have 64 lanes on the enthusiast platform in a year or two!

[patrickjp93]

The issue is if you don't shrink the die then you can't add more resistors without changing the architecture. And optimizing the architecture for the specific die size is harder and harder to do each time you improve it. 

 

What'll end up happening, inevitably after 2025, is they will start increasing the component size to add in more transitions because it'll be the only way to increase power but obviously heat output, power consumption, and physical size will all suffer because of that. 

Nope, CNTs will take us to 2025 with die shrinks every 2 years down to .300 nm, and graphene can take us to 2035. At that point quantum is the only place left to go. Also, optimizing the architecture is easier. As electrical paths shorten IPC goes up, hence Broadwell shaving 2 cycles off the FPU mul/div ops. You keep the least used instructions on the outside of the chip and the most used towards the center.

[bomerr]

Nope, CNTs will take us to 2025 with die shrinks every 2 years down to .300 nm, and graphene can take us to 2035. At that point quantum is the only place left to go. Also, optimizing the architecture is easier. As electrical paths shorten IPC goes up, hence Broadwell shaving 2 cycles off the FPU mul/div ops. You keep the least used instructions on the outside of the chip and the most used towards the center.

http://www.extremetech.com/computing/165331-intels-former-chief-architect-moores-law-will-be-dead-within-a-decade

 

Quote "Innovation is still going to happen. There are technologies that are going to continue to improve our underlying level of ability; a 30x advance in 50 years is still significant. But the old way — the old promise — of a perpetually improving technology stretching into infinity? That’s gone. And no one seriously thinks graphene, III-V semiconductors, or carbon nanotubes are going to bring it back, even if those technologies eventually become common."

 

and from the comments 

 

"At 14 nm, fewer than 30 Si atoms span the channel and quantum tunneling is significant. It will be far worse at each successive node. Ultimately, at about 2.5 nm, the channel will be spanned by just 5 Si atoms and the Heisenberg uncertainty principle tells us that it will no longer be possible to know that an electron is contained within a single trace"

 

​Intel's roadmap ends at 5nm because thats kinda it. We have 4 die shrinks and 8 processor architectures left then its all over. After that we are going to be in the land of incremental upgrades where it's uncertain how and when the next breakthrough/leaps happens. AND THAT ISN'T GOOD FOR BUSINESS which is why it doesn't make any sense for Intel to skip broadwell. 

[patrickjp93]

http://www.extremetech.com/computing/165331-intels-former-chief-architect-moores-law-will-be-dead-within-a-decade

 

Quote "Innovation is still going to happen. There are technologies that are going to continue to improve our underlying level of ability; a 30x advance in 50 years is still significant. But the old way — the old promise — of a perpetually improving technology stretching into infinity? That’s gone. And no one seriously thinks graphene, III-V semiconductors, or carbon nanotubes are going to bring it back, even if those technologies eventually become common."

 

and from the comments 

 

"At 14 nm, fewer than 30 Si atoms span the channel and quantum tunneling is significant. It will be far worse at each successive node. Ultimately, at about 2.5 nm, the channel will be spanned by just 5 Si atoms and the Heisenberg uncertainty principle tells us that it will no longer be possible to know that an electron is contained within a single trace"

 

​Intel's roadmap ends at 5nm because thats kinda it. We have 4 die shrinks and 8 processor architectures left then its all over. After that we are going to be in the land of incremental upgrades where it's uncertain how and when the next breakthrough/leaps happens. AND THAT ISN'T GOOD FOR BUSINESS which is why it doesn't make any sense for Intel to skip broadwell. 

Intel's roadmap goes to 3nm and IBM's goes beyond. There's a reason IBM left the standards communities. They're all too narrow-minded.

[ChrisxIxCross]

DDR3/DDR4... So multiple skus of motherboards with different ram compatibility? Not very smart imo. Ive seen people already buy X99 boards w/ DDR3 ram lol.

[DigitalHermit]

Aww crap... I might have to skip out on 2 Intel generations if I plan on going Intel long-term...

 

If they're skipping Broadwell which was supposed to be compatible with Z97, I might have to drop the G3258 rig...

 

Looks like 2016 might be the sweet year of building something...

 

(I just got handed down a C2Q system... might have to live with that for a while...)

[kacper]

Could anyone explain to me how do you think quantum computing is supposed to replace convecional computing ?

Last time I've had quantum computing classes  , anyone knew that QPU's suck ass when it comes to regular CPU tasks.

Unless there will be a breakthrough in quantum programming , there's no chance that processing unit using only QM will replace CPU's . Hybrids? More likely