• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Kaby Lake information.

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
jpiniero said:
It is very unlikely that Cannonlake would support the same socket.
Click to expand...

Things are a little goofy with Skylake, but wouldn't traditionally Cannonlake use the LGA1151 socket as the die shrink of the SL uarch?

Of course, keeping the same socket it kind of pointless from the user perspective if the chipsets make an upgrade mandatory anyway to move from SL to KL or CL.
 
MrTeal said:
Things are a little goofy with Skylake, but wouldn't traditionally Cannonlake use the LGA1151 socket as the die shrink of the SL uarch?
Click to expand...

Traditionally, yes, but these times are anything from traditional. Considering what Intel did to Skylake to sorta get it down to tablet levels you can only imagine what they've done to Cannonlake.
 
III-V said:
I think you messed up the wording on your question. Skylake will work on KL motherboards -- 4+4e Skylake will not work on today's Skylake motherboards, though. I am not positive as to why, but yeah.
Click to expand...

IlllI said:
you're right that I am confused hah. I've never heard of a processor not compatible with a future motherboard, but the future cpu will be compatible with a past motherboard. I assume they are pin compatible for that to be possible, but wouldn't that confuse a lot of people?
Click to expand...

The answer is power and can be found in this June 2014 WCCF post (it seems this time they had accurate information).

Look at the power requirements of DT 4+4e compared to 4+2.
4+4e: VccGT requires 4 phases and a GTUS rail
4+2: VccGT requires 2 phases and no GTUS rail.

For a measure of confirmation, Intel's Skylake-S datasheet, page 113 lists a VccGTX rail, which is only connected for processors with GT3/4.

Something like 100% of existing boards do not meet the 4+4e power requirement, so nobody can get 4+4e without a new motherboard.

Given 4+4e is coming out at the same time as the 200 series chipset, Intel thought it pointless to spend the time/effort to validate against the older 100-series chipset when nobody would take advantage of it.
 
Techhog said:
This is the most egregious example of revisionist history I've read in a while... Well, unless you back it up.
Click to expand...

You can start with this, that you may most likely reject. Dothan to Broadwell. 2003 to 2015.

intel-xeon-ipc-chart.jpg

http://www.nextplatform.com/2015/03/09/intel-crafts-broadwell-xeon-d-for-hyperscale/

Else some good information in this post:
http://forums.anandtech.com/showpost.php?p=37939597&postcount=106
 
Last edited:
So then has there been any type of official/credible leak or something put out by Intel that Cannonlake (I know this is a Kaby Lake thread) is going to be able to work in Kaby Lake boards? Because if that's the indication, I may just put off my build until Kaby Lake is available.

I wonder if Intel has hit the point on the desktop of just not needing a new chipset/socket for future CPU tech...?
 
Cannonlake would have to use its own single usage socket if it is so. Because Icelake will use a new one with the reintroduction of the FIVR.

Cannonlake have also always been planned for LGA1151.
 
I think we will see a new chipset with Cannonlake. Jury's out on whether it will come w/ a new socket, but I'd bet on "yes" if only to give Intel's motherboard partners a break 🙂
 
frozentundra123456 said:
Well, the tocks have been 10+ % gains, not 5% like the thread you were replying to.
Click to expand...

The poster claimed everything after SB was 5%. Its not. Both Haswell and Skylake is above that.

Some people have a completely wrong notion that before Sandy Bridge, IPC gains was wild and crazy. But you have to go all the way back to Pentium Pro for that to be true.
 
ShintaiDK said:
The poster claimed everything after SB was 5%. Its not. Both Haswell and Skylake is above that.

Some people have a completely wrong notion that before Sandy Bridge, IPC gains was wild and crazy. But you have to go all the way back to Pentium Pro for that to be true.
Click to expand...

Most people generally include the IPC regressions through Netburst and the massive jump with Core 2, since those were Intel's primary product lines for a half decade on desktop and server.
 
mikk said:
Can you explain why? I have to disagree, it is very unlikely imho that CNL will get a new socket. Intel don't change sockets for a refresh generation.
Click to expand...

Kaby Lake is the refresh generation. Think of it as the tock. You're getting the 200 series chipset and a bunch of new features like 3DXP support. Intel seems like it only really does the socket compatibility in the first place just to make the lives of OEMs a little bit easier to transition from one processor to the second one but I'm not even sure they care now given that laptops are now all BGA and they don't seem to have a problem with it. But they limit it to two gens.

Cannonlake is also rumored to be an SoC with the PCH ondie. They could have came up with some sort of scheme to disable the PCH to keep BC but it obviously doesn't make sense now with Kaby Lake in the fold. Maybe it will have FIVR too.
 
ShintaiDK said:
Cannonlake would have to use its own single usage socket if it is so. Because Icelake will use a new one with the reintroduction of the FIVR.

Cannonlake have also always been planned for LGA1151.
Click to expand...

They were stuck on this node for 3 generations though, and it's likely only going to continue to get worse. Which means they would likely get 3 or more generations out of a new socket 10nm socket as well, which should be sufficient to merit the development cost.

Intel is known for switching sockets fairly frequently too.

Then again, a new socket might mean it lowers CPU sales due to the motherboard requirement.

Of course, they sell motherboards too.

EDIT: Where did you find info on icelake? The only thing I see is an article from july that states Intel Ditches Cannonlake And Replaces It With Ice Lake CPUs.

"Skylake and Kaby Lake chips will be compatible with LGA1151 socket motherboards, the die shrink should mean a new form-factor as well as new mainboards."
 
Last edited:
mysticjbyrd said:
They were stuck on this node for 3 generations though, and it's likely only going to continue to get worse. Which means they would likely get 3 or more generations out of a new socket 10nm socket as well, which should be sufficient to merit the development cost.

Intel is known for switching sockets fairly frequently too.

Then again, a new socket might mean it lowers CPU sales due to the motherboard requirement.

Of course, they sell motherboards too.

EDIT: Where did you find info on icelake? The only thing I see is an article from july that states Intel Ditches Cannonlake And Replaces It With Ice Lake CPUs.

"Skylake and Kaby Lake chips will be compatible with LGA1151 socket motherboards, the die shrink should mean a new form-factor as well as new mainboards."
Click to expand...

There is no ditching of Cannonlake. Icelake follows Cannonlake and will reintroduce the FIVR.
 
ShintaiDK said:
There is no ditching of Cannonlake. Icelake follows Cannonlake and will reintroduce the FIVR.
Click to expand...
Replacing or at least renaming Cannonlake with Icelake would make a lot of sense. Per Intel's recent naming convention another Lake part would be the same u arch introduced in Skylake, and four generations of that seems unlikely.
 
Last edited:
jpiniero said:
Kaby Lake is the refresh generation.
Click to expand...

Cannonlake as well.

jpiniero said:
Think of it as the tock.
Click to expand...

It isn't.

jpiniero said:
You're getting the 200 series chipset and a bunch of new features like 3DXP support.
Click to expand...

Chipset and socket are two different things. I don't see how this can be a proof that Intel will be using a different socket for Cannonlake. There is absolutely no indication for a new socket. Intel could launch new chipsets whenever they want, this isn't connected to a new socket. Z97 wasn't for example.


jpiniero said:
Cannonlake is also rumored to be an SoC with the PCH ondie.
Click to expand...

What is your source?
 
mikk said:
Chipset and socket are two different things. I don't see how this can be a proof that Intel will be using a different socket for Cannonlake. There is absolutely no indication for a new socket. Intel could launch new chipsets whenever they want, this isn't connected to a new socket. Z97 wasn't for example.
Click to expand...

Well, I don't have proof... that's why I said "very unlikely" and not "not happening". But it sure looks like Intel limits to two gens to force OEMs to upgrade motherboards and to limit the need to worry about BC, do you not agree?

Plus, Intel had to remove FIVR from Skylake. I'm sure they would love to get it back in ASAP. The upside of keeping BC with 100/200 series just doesn't seem like it's worthwhile even if Intel does end up selling Cannonlake models without the PCH.

What is your source?
Click to expand...

http://www.itworld.com/article/2988...generation-could-be-going-to-eight-cores.html

Not the best source of course, but it's been rumored for awhile. It makes sense since based upon what Intel has been doing the #1 priority for Cannonlake is getting Core into phones... and for that you really need a SoC.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top