Intel's 22nm Successors Revealed - Ivy Bridge & Haswell

[y2kbos]

ignorant is what it is.

Maybe.

Larabee is at least an example of them thinking outside the box.Im not an engineer or anything, but die shrinks, a few more cores and cache isnt gonna make me clean that sock when I am done with it

[nfm]

Still no tri-gate transistors, they were supposed to come in 2007 .

[Shintai]

Maybe.

Larabee is at least an example of them thinking outside the box.Im not an engineer or anything, but die shrinks, a few more cores and cache isnt gonna make me clean that sock when I am done with it

Larabee is still something that is developed for what...4-5 years aswell? And its not really thinking outside the box. Its just a reintroduction of the software renderer we had before and under the Voodoo times. Its things done right however. Sicne GPUs sofar is more jump over where the fence is lowest.

And Larabee is basicly Die shrink, more cores and more cache......

[Helmore]

I prefer ticks.the diference between the 65nm tock and the conroe tick was significant. As is the difference between the 45nm tock and the nehalem tick.

the srinks just didn't do too much though. 10% higher clocks, <10% more performance/clock... <20% performance boosts don't impress me as much as 20-100% performance boosts.

You are switching the ticks and tocks arounds. Conroe was a tock, while Penryn was a tick and Nehalem is a tock again. So in essence you agree with me

[Kingcarcas]

^ Hehe, yah everyone gets them mixed up. Westmere will be here next year, hopefully DDR3 is cheap, mobos matured, if i need to upgrade. 6 Cores with HT? Sounds tasty

[CraptacularOne]

You are switching the ticks and tocks arounds. Conroe was a tock, while Penryn was a tick and Nehalem is a tock again. So in essence you agree with me

Actually if you are going by the core architecture, Conroe was the "tick" and Penryn was the "tock". They both are on the core 2 architecture with the Penryn family benefiting from a smaller die size and the SSE4.1 instruction set and a few other minor enhancements.

Nehalem will start this cycle over again. Making it the "tick" and Sandy Bridge the "tock".

[Helmore]

You could do it that way, but Intel has been using the "tick" for introducing a new process technology and the "tock" is a newer CPU on that same, but more mature, process technology. All I'm doing is going by what Intel uses to explain their strategy....

But that's all nitpicking and does not really matter in the end. As long as we all get what was meant to be said.

[y2kbos]

Larabee is still something that is developed for what...4-5 years aswell? And its not really thinking outside the box. Its just a reintroduction of the software renderer we had before and under the Voodoo times. Its things done right however. Sicne GPUs sofar is more jump over where the fence is lowest.

And Larabee is basicly Die shrink, more cores and more cache......

True, I just dont find these latest plans all that exciting, maybe I am growing out of it who knows.I kind of want more you know?

[Speederlander]

I dont like it, as nothing should be set in stone.

Roadmaps have no place in Tech, especially 4 year roadmaps.Tech should be spontaneous.

Yeah, it's been just disasterous for Intel. Hasn't worked well at in getting powerful and exciting new technologies to the market. Far better to embrace a model like AMD's, where "planning" and "road-maps" are apparently four-letter words.

[STaRGaZeR]

Actually if you are going by the core architecture, Conroe was the "tick" and Penryn was the "tock". They both are on the core 2 architecture with the Penryn family benefiting from a smaller die size and the SSE4.1 instruction set and a few other minor enhancements.

Nehalem will start this cycle over again. Making it the "tick" and Sandy Bridge the "tock".

Is the other way around. New architecture = tock. Die shrink = tick. Conroe was the tock, Penryn is the tick. And Nehalem is the next tock.

Tick tock madness.

[CraptacularOne]

Is the other way around. New architecture = tock. Die shrink = tick. Conroe was the tock, Penryn is the tick. And Nehalem is the next tock.

Tick tock madness.

I am merely explaining a point of view. Either is correct, as it really is irrelevant anyway.

[xlink]

ehh I guess I did mix them. all I know is that I heard of tick-tock right around conroe time, so conroe got to be the tick in my mind.

[Kingcarcas]

Hehe, from my understanding the Tick is the shrink, Pentium 4 and D had a 65nm shrink, then you have the new arch. on that process AKA Conroe or Tock.

[fragmasterMax]

I have to agree with what y2kboss was saying about thinking outside the box. In 2009 6!!! years after k8, amd is finally releasing a new ground up cpu architecture (Bulldozer). (how many has intel been through since then (mobile celeron, p4, core duo, atom, etc) They had a good thing going, but they should have anticipated Intel with it's superiority consumer base, and clock speed would eventually use it's superior revenue to make something much better.

Still, it is very ignorant to think that this stuff can just be built in a few months or something planning is a must, and intel has got a great future.

[EternityZX9]

I dont like it, as nothing should be set in stone.

Roadmaps have no place in Tech, especially 4 year roadmaps.Tech should be spontaneous.

*Hands award for most-outlandish-remark-ever-made-in-a-post-on-XS-before*

[terrace215]

I have to agree with what y2kboss was saying about thinking outside the box. In 2009 6!!! years after k8, amd is finally releasing a new ground up cpu architecture (Bulldozer). (how many has intel been through since then (mobile celeron, p4, core duo, atom, etc) They had a good thing going, but they should have anticipated Intel with it's superiority consumer base, and clock speed would eventually use it's superior revenue to make something much better.

Still, it is very ignorant to think that this stuff can just be built in a few months or something planning is a must, and intel has got a great future.

There's no 2009 Bulldozer anymore. 2009 and 2010 H1 all use K8L, basically. Server OEMs forced AMD to get more realistic with their roadmaps. Of course, I wouldn't put it past AMD to "rename" Magny-Cours to "bulldozer", but it isn't the ground-up architecture Bulldozer that made a brief appearance in July 2007 tech day slides.

[JumpingJack]

Says who?

The past is the past.

CPU designers ...
http://ieeexplore.ieee.org/Xplore/lo...arnumber=63355

Quote: "4 years to design the 386 from start to first silicon" - you will need a library to verify it.


http://www.computer.org/portal/site/...3g1!1369782199

"The P6 project would eventually grow to over 400 design and validation engineers and take 4.5 years to production. But that huge investment paid off—P6 became the Pentium Pro microprocessor, was adapted to become the Pentium II and then the Pentium III, and, most recently, has evolved into the Centrino mobile line. "

I can't find any references to AMD's or IBM's design time, you could try to look here:

http://www.research.ibm.com/journal/rd51-6.html

I would think this is fairly accurate.

[saaya]

sounds like haswell will be soc or at least sop?
interesting...

does anybody know amds plans?
do they even still make plans for 2011? :P

[strange|ife]

damn 22nm by 2011? hmm..

getting close to the theoretical limits of Silicon. Read somewhere down past 9nm we would need to move to a whole new process and substrates, or just say 128 9nm cores on a die. quantum computing etc. only so far they can go with microprocessors

[mAJORD]

I am merely explaining a point of view. Either is correct, as it really is irrelevant anyway.

haha no it's not.

http://www.intel.com/technology/tick-tock/index.htm

Tock = New architecture

Tick = Shrink.

Very simple, very spelled out, for the love of god everyone memorise it, nothing looks more stupid, and 'page wasting' than a tick tock argument.

[savantu]

What do you think they were doing? They were planning the Core 2 architecture a few years ago. It's not like the Core 2 architecture was a "spontaneous" development, where one R&D guy just shouted "EUREKA!" Then they rushed off to develop the Core 2 line. They spent years developing the technology.
...

Don't tell him Core development started in early 2002 and Conroe was released in mid 2006.

Basically , they started work on Core shortly after Northwood was released.

It's spontaneous for sure...

[largon]

does anybody know amds plans?
do they even still make plans for 2011?

I hope we will know after AMD Tech Day '08 on November 13th.

[saaya]

damn 22nm by 2011? hmm..

getting close to the theoretical limits of Silicon. Read somewhere down past 9nm we would need to move to a whole new process and substrates, or just say 128 9nm cores on a die. quantum computing etc. only so far they can go with microprocessors

hahah dont listen to that BS man...
when i started to get into pcs they said anything past 65nm is gonna be impossible, then later they said anything past 45nm will be really difficult/impossible, then they said anything past 32nm will be really difficult/impossible...

thats just technology... from where your standing now you can only see so far into the future of whats coming next. after a certain point its just purely fantasy, some people are pessimistic and think there wont be any room for improvements, others are overly optimistic and dream about flying cars and opening wormholes :P

leakage was supposed to be the ultimate issue for shrinking past 65nm, and we are not at 45nm and beyond and leakage is actually better than on 65nm and 90nm so... just dont listen to those rumors :P

there are plenty of smart heads on this planet who figure out how to solve the problems we MIGHT run into in the next 5 years

[saaya]

I hope we will know after AMD Tech Day '08 on November 13th.

i hope they finally woke up and have some actual ideas what they can do NOW and in short term future and stop dreaming about some distant future product thats gonna be soooo awesome... they need to stop recycling their fusion and other older roadmap bs :P

if amd doesnt get a serious adrenaline rush now its gonna be intel and nothing but intel for the next decade or two... maybe longer...

and dont get me wrong, intel has changed a lot and they are doing really great stuff now and listening to their customers, support the enthusiasts and all... but without competition even the nicest company with the nicest ceo will end up with world domination plans :P

[Shintai]

hahah dont listen to that BS man...
when i started to get into pcs they said anything past 65nm is gonna be impossible, then later they said anything past 45nm will be really difficult/impossible, then they said anything past 32nm will be really difficult/impossible...

thats just technology... from where your standing now you can only see so far into the future of whats coming next. after a certain point its just purely fantasy, some people are pessimistic and think there wont be any room for improvements, others are overly optimistic and dream about flying cars and opening wormholes :P

leakage was supposed to be the ultimate issue for shrinking past 65nm, and we are not at 45nm and beyond and leakage is actually better than on 65nm and 90nm so... just dont listen to those rumors :P

there are plenty of smart heads on this planet who figure out how to solve the problems we MIGHT run into in the next 5 years

We still have the atomic size issue. So at some point in our lifetime we will hit the wall. With this pace, maybe before 2020.