Discussion Intel current and future Lakes & Rapids thread

[StinkyPinky]

People who buy Z370 know this beforehand now, so it's on them. But yes, it's the downside of a crammed in between release.

Perhaps, but I find it quite unethical that a cpu released in October won't work in motherboards released in the same year, and then to top it off the motherboard it will work in won't be compatible with the next cpus released less than 12 months later.

 

[Ajay]

Already posted by Sweepr in the other thread. However, since this is the everything after Coffee Lake thread I thought I'd post it here as well for future reference.


Intel Z390 to support 8C/16T CPUs in H2 2018

https://videocardz.com/newz/intel-z390-to-support-8c16t-cpus-in-h2-2018

What??? Is Ice Lake due in 2H18?!

Edit: I guess Q4 does make sense, after thinking about it..

 

[Jan Olšan]

It would be pretty awesome.

There is an alternative explanation: the Eurocom public/social communications dude got confused by the old linkedin post that states eight cores for Cannon Lake and he doesn't know that Cannon Lake-S was cancelled.
Remember that Intel calls Z390 "Cannon Lake PCH". That could make somebody think that Z390 would still see a release of the originally planned Cannon Lake octocores, and the H2 2018 date could be just guess. This way, the much unexpected compatibility with Z390 and presumably LGA 1151 would be explained, too.

I have no idea what to believe here though. I could see how somebody could fall for this mistake (and this long in advance, the laptop maker is not allowed to talk NDA roadmap of course). But on the other hand, it is a complicated theory. Really, who knows, it might actually be Ice Lake for real.

 

[DrMrLordX]

It could be very good for Intel to release Icelake in Q4 2018. Or sooner. If they can meet that schedule, it would say a lot about their ability to fix the 10nm mess, assuming Icelake isn't also going to be 14nm++. Though if 14nm++ has better performance, maybe Intel would be better off just using that for awhile . . .

 

[mikk]

There probably will be a 400 series chipset, just no Z.

Edit: Intel could in theory for instance release Icelake-K in Oct 2018 with Z390 and then release the rest of the Icelake-S lineup in Jan 2019 with B/Q/H 400 series.

If there is a 400 series with a new chipset you can be sure there is a Z board as well, it makes no sense otherwise.

 

[Ajay]

If there is a 400 series with a new chipset you can be sure there is a Z board as well, it makes no sense otherwise.

Probably, but Intel isn't making a whole ton of sense lately with the delays @ 14nm & 10nm. I hope they get their sh*t back together soon.

 

[beginner99]

So basically to rush out coffeelake-k 3 month earlier they needed to make the z370 chipset for whatever reason. Since they want to get that cost back in, the real z370, now called z390, gets delayed roughly 6 months. This is idiotic. They should have just waited till chipset is ready...

 

[VirtualLarry]

So basically to rush out coffeelake-k 3 month earlier they needed to make the z370 chipset for whatever reason. Since they want to get that cost back in, the real z370, now called z390, gets delayed roughly 6 months. This is idiotic. They should have just waited till chipset is ready...

AMD's really got Intel scramblin'...

 

[Arachnotronic]

So basically to rush out coffeelake-k 3 month earlier they needed to make the z370 chipset for whatever reason. Since they want to get that cost back in, the real z370, now called z390, gets delayed roughly 6 months. This is idiotic. They should have just waited till chipset is ready...

So what you're saying is you want Intel to hold back delivering better CPUs because...why, again?

 

[osgorth]

So basically to rush out coffeelake-k 3 month earlier they needed to make the z370 chipset for whatever reason. Since they want to get that cost back in, the real z370, now called z390, gets delayed roughly 6 months. This is idiotic. They should have just waited till chipset is ready...

And miss the entire holiday season? That's hardly an option.

I'm certain they're doing this to be able to sell Coffee Lakes during all of Q4, and also to reclam the main desktop performance crown and scare AMD back into the budget corner.

 

[beginner99]

So what you're saying is you want Intel to hold back delivering better CPUs because...why, again?

Because I now have the choice to wait till summer for z390 that is compatible with icelake or chose z370 as a dead end platform now. This wasn't really a huge issue from SB to SKL. It was know that next core would be a quad as well and offer minor IPC improvements. So you could get it and knew that for the next 2+ years there would not be anything worth upgrading to in same price segment. Now I have to chose whether I want the option to later upgrade to an octo-core or not and if I want that option I have to wait 6 month longer than if they just released Z390 in January.

And people now are complaining about z270 and no CFL. In 1 year they will complain about Z370 and no Ice lake.

 

[Glo.]

If Z390 is compatible with both Coffee Lake and Ice Lake I may wait for this chipset with my purchase. At the time of release also should be released Volta GV107 chip, to fit in perfectly.

 

[LTC8K6]

AMD's really got Intel scramblin'...

Reminds of the (likely bogus) Yamamoto quote about awakening sleeping giants.

 

[Bouowmx]

https://newsroom.intel.com/news/intel-technology-manufacturing-day-china/

Intel reveals a wafer with 10 nm Cannon Lake, and another with ARM Cortex-A75 (> 3.0 GHz).

 

[LTC8K6]

https://newsroom.intel.com/news/intel-technology-manufacturing-day-china/

Intel reveals a wafer with 10 nm Cannon Lake, and another with ARM Cortex-A75 (> 3.0 GHz).

The giant stirs...

 

[Dayman1225]

Wow! 3ghz, isn't that quite high for ARM SoCs?

 

[Lodix]

ARM designed Cortex A75 to run at 3GHz using TSMC's 10nm with low TDP. So I don't see this as something promising.

 

[Arachnotronic]

ARM designed Cortex A75 to run at 3GHz using TSMC's 10nm with low TDP. So I don't see this as something promising.

Have you seen any high frequency TSMC or Samsung 10nm implementations?

 

[Lodix]

Have you seen any high frequency TSMC or Samsung 10nm implementations?

Don't blame foundries for a bad implementation of the chip designer. And in Samsung case they had terrible yields. ARM tests their desings at TSMC factory. Even GF's copy of 14nm can reach over 3GHz. Also this year they focused in lowering power consumption of the chips overall.

 

[Dayman1225]

https://newsroom.intel.com/newsroom/wp-content/uploads/sites/11/2017/09/falcon-mesa-intel-tmg-china-event-fact-sheet.pdf


Intel has released some first details of its 10nm FPGA. Will use 2nd gen EMIB and next-gen HBM.

It's not directly related, Intel seem to be announcing/teasing quite a bit of 10nm products.

Cannonlake
Icelake
Tigerlake

Possible Intel 10nm arm cores
Falcon Mesa FPGA

10nm looking on the up?

 

[NTMBK]

https://newsroom.intel.com/newsroom/wp-content/uploads/sites/11/2017/09/falcon-mesa-intel-tmg-china-event-fact-sheet.pdf


Intel has released some first details of its 10nm FPGA. Will use 2nd gen EMIB and next-gen HBM.

It's not directly related, Intel seem to be announcing/teasing quite a bit of 10nm products.

Cannonlake
Icelake
Tigerlake

Possible Intel 10nm arm cores
Falcon Mesa FPGA

10nm looking on the up?

FPGAs have very high prices, meaning they have big enough profit margins to handle crummy yields. I'll be more impressed when they start shipping consumer parts in volume, as that will mean that they've fixed the yield problems.

 

[Dayman1225]

With this happening I assume something is on the up.... or I hope, from what I hear this is usually released 1~ year before launch so, H2 2018 Q3/4.

Keyword : Hope

 

[Ajay]

I wonder if Ice Lake will be using PCIe 4, at least for the DMI connection.

 

[Arachnotronic]

I wonder if Ice Lake will be using PCIe 4, at least for the DMI connection.

Well, it's getting a new PCH, so it's at least possible.

 

[Dayman1225]

Cannon lake wafer

Intel 10nm ARM Wafer

https://www.anandtech.com/show/11850/intel-displays-10nm-wafer-commits-to-10nm-falcon-mesa-fpgas?