Intel "Coffee Lake" Architecture by Q2-2018, 7 nm Process By 2022?

[alucasa]

I bet internal code named for Coffee lake is f--king cake. Engineers just slack while making minimal improvements and just get paychecks and fu- cakes.

 

[INSTG8R]

AMD at this moment can only catch Intel on the price level...

Zen performance is still just speculation much like this comment....

 

[cdawall]

...thinking that a CPU you bought in 2020, is only 40% faster than the one you bought in 2007...

Having owned a multitude of C2D based stuff and a current intel...I'm trying to figure out 40% in what? There is literally no benchmark that shows any less than 100% clock for clock improvements and sometimes 150-200% especially when multithreading is taken into account.

 

[dorsetknob]

will they Decaffeinate these Cpu's for a Celeron/Pentium SKU ??
sorry had to try to be differently Funny
"My brain hurts now"

 

[BiggieShady]

All of you saying code name is stupid shut up, it's great. People have strong emotions towards coffee just as to cpus, and they want their cpus to run like they've drunk a lake of coffee ... and by that I don't mean to the bathroom

 

[MxPhenom 216]

After my recent rig mishap I was contemplating going to Kaby Lake in February next year, but now that I have a z97 board and could run a m.2 nvme drive if I wanted to, no reason now other than maybe to have a processor that overclocks better, but eh. Ill stick with my Haswell platform for probably another 2 years and then go all new. By that time ill be done with school too, and maybe even getting Intel chips free or at a discount (assuming I get a EE job there )

 

[yogurt_21]

Having owned a multitude of C2D based stuff and a current intel...I'm trying to figure out 40% in what? There is literally no benchmark that shows any less than 100% clock for clock improvements and sometimes 150-200% especially when multithreading is taken into account.

baseclock alone would account for 40% Q6700 (released in 2007) being 2.66Ghz no turbo compared with the i7 7700K 3.6Ghz with a 4.2Ghz turbo both at a similarish price point (the Q6700 debuting at $530 US).

Then you factor in hyperthreading, instruction improvements, other architectural improvements, memory bandwidth,chipset improvements, and etc.

The difference could be higher than 500% in certain areas like video encoding, compression, etc. and that's just 2007- 2016 much less what will be out in 2020.

 

[cdawall]

are you a retard?

Are you arguing he is incorrect?



262 CB points for a stock QX6700, there are mobile 3rd gen mobile i7 quads showing 100% scaling per clock

 

[Durvelle27]

Coffee Lake

 

[Prima.Vera]

Guys chill! I meant Sandybridge CPU. I thought was released in 2007. Is actually 2011. My bad.
Now let's move on.

 

[FordGT90Concept]

Moore's Law...

This is horrifying for tech. Without continually improved process tech, not only will the processor industry stagnant, but all subsequent industries that rely on better, cheaper processors will too.

 

[tabascosauz]

This rumor article has left me very confused. Weren't there multiple confirmations over the past year+ that Cannonlake would be the successor to Kaby Lake, and not some low power family relegated to -U SKUs?

The original understanding of 10nm Cannonlake very much aligns with PROCESS - ARCHITECTURE - OPTIMIZATION that has been introduced with Kaby. I thought Icelake and Tigerlake were to eventually round out the PAO approach, all being on 10nm (though the HotHardware article gives the three CPU families different fancy names, at the end of the day 10nm 10nm+ and 10nm++++++++ are all 10nm).

And thus as the 'lakes round out 10nm, that should bring us to 2021/2022 with 7nm. Was this not how is was expected before?

I don't think some people understand the difference between the physical difficulties of going under 10nm and the "milking" that Intel is doing in the absence of AMD's competition. These major fabs don't adopt the exact same process, and I wouldn't be surprised that assuming GloFo actually delivers on its 7nm process (GloFo hasn't the greatest reputation of delivering what is agreed), it would be something like how "20nm" HPM in the ARM space wasn't exactly "20nm" in all respects because of the nature of the process.

 

[Frick]

So time for bigger chips then?

 

[zelnep]

Are you arguing he is incorrect?



262 CB points for a stock QX6700, there are mobile 3rd gen mobile i7 quads showing 100% scaling per clock

why are you comparing good ol' cpu when there was competition to current intel cpus? by the way - without HT in HT heavy benchmark - "nice one"... how about compare i7-2700k vs the "latest and greatest" i7-6700k - there will not be even +25% - 4 gens 5 years and not even +25%.

 

[1c3d0g]

Just watch those roadblocks crumble away if AMD puts pressure on revenues with Zen sales.

Don't be delusional. Everybody is bumping into the same physics problems. Intel, AMD, TSMC, Samsung, GlobalFoundries, IBM, you name it, the physics don't change!!!

Could they (Intel) lower prices? Yes, they could. But thinking about doubling performance in the next few years or some other crazy sh!t like that is wishful thinking at best, without some new exotic material at the very least. Let's face it, silicon is nearing the end-of-life as a product. It has served us all well for the past decades, but the time will come that it has to be put to rest (and that time is closer than most people think). Hopefully the smart guys can crack this bastard with a new type of material before we get stuck, technologically speaking.

 

[cdawall]

why are you comparing good ol' cpu when there was competition to current intel cpus? by the way - without HT in HT heavy benchmark - "nice one"... how about compare i7-2700k vs the "latest and greatest" i7-6700k - there will not be even +25% - 4 gens 5 years and not even +25%.

When the second gen of the I series comes out in 2007 it will be pertinent to what we were talking about.