[Updated, confirmed] Intel Coffee Lake CPUs will NOT be compatible with Z270 chipset motherboards

[Drak3]

Just now, DeadEyePsycho said:

Devil's Canyon was not a new architecture. It's literally just like 2600K and 2700K, just they gave it a code name.

Scrolling back up is hard.

38 minutes ago, Drak3 said:

if we're being honest, Devil's Canyon should not be considered anything more that a mid product fix, as Intel doesn't consider it enough to be a different generation than Haswell, even with their minimal requirements.

 

[Grinners]

1 hour ago, MidnightBanshi said:

*sigh*.....built my first gaming/editing computer, thinking it was going to last a while............NOPE!

Looks like it'll be either Ryzen 7 or ThreadRipper 1900X for me after this one.  So long Intel.  It was a short ride with a jolt near the end.

It is annoying that computers instantly stop working the moment newer models/versions come out... /s

[2FA]

15 minutes ago, Drak3 said:

Scrolling back up is hard.

 

Nope I saw that, and then you included DC after the fact. It was still considered 4th generation, Skylake is 6th and Kaby considered 7th. Intel hasn't supported more than two generations on a socket in a long time.

[Drak3]

Just now, DeadEyePsycho said:

Nope I saw that, and then you included DC after the fact. It was still considered 4th generation, Skylake is 6th and Kaby considered 7th. Intel hasn't supported more than two generations on a socket in a long time.

Kabylake also is not a new architecture. It's a (small) refinement of Skylake. Just like DC was to Haswell.

But beyond this comment, the argument isn't worth my time, especially when the context of the conversation was Intel doing 2 product lines per socket, when the previous most technically had 3.

[2FA]

Just now, Drak3 said:

Kabylake also is not a new architecture. It's a (small) refinement of Skylake. Just like DC was to Haswell.

But beyond this comment, the argument isn't worth my time, especially when the context of the conversation was Intel doing 2 product lines per socket, when the previous most technically had 3.

The word you're looking for is generation, not product line. A product line would be Core, Xeon, Atom, etc.

[Morgan MLGman]

3 hours ago, DeadEyePsycho said:

Nope I saw that, and then you included DC after the fact. It was still considered 4th generation, Skylake is 6th and Kaby considered 7th. Intel hasn't supported more than two generations on a socket in a long time.

Still, Kaby Lake to Skylake is what Devil's Canyon to Haswell was - no IPC improvements, just minor things including a new pointless chipset and higher average OCs.

People (including me) are mad because it's possible that it's Intel's artificial limitation that blocks previous LGA1151 chipsets from using Coffee Lake, created so people buy more motherboards. Why is the socket the same if previous chipsets don't fit? I've got no idea, but it was a retarded decision.

[Jito463]

8 hours ago, RxL said:

so please can somebody inform why any of you are even mad / upset about this news of coffee lake not being compatible with z270

The problem is that the 200 series chipset is less than a year old (barely over 6 months, in fact), and is already a dead end line.  If the 200's and the 7th gen CPUs had launched beginning of last year, it wouldn't be as big of an issue.  Sure, there would still be some people upset, but at least it would have had a decent life cycle.  Less than one year for a new chipset is just way too short.

[mr moose]

4 hours ago, Morgan MLGman said:

Still, Kaby Lake to Skylake is what Devil's Canyon to Haswell was - no IPC improvements, just minor things including a new pointless chipset and higher average OCs.

People (including me) are mad because it's possible that it's Intel's artificial limitation that blocks previous LGA1151 chipsets from using Coffee Lake, created so people buy more motherboards. Why is the socket the same if previous chipsets don't fit? I've got no idea, but it was a retarded decision.

Several reasons have been postulated that aren't "retarded",  but given so far it is only a tweet and no official statements it is possible that this is all just BS.

 

[Morgan MLGman]

2 minutes ago, mr moose said:

Several reasons have been postulated that aren't "retarded",  but given so far it is only a tweet and no official statements it is possible that this is all just BS.

 

I just had a thought: "I wonder how many people would've bought Z270 platform if Intel announced that next-gen CPUs released around a year later won't be supported in those mobos"

 

This is the problem with Intel and this is what I really don't like.

[mr moose]

Just now, Morgan MLGman said:

I just had a thought: "I wonder how many people would've bought Z270 platform if Intel announced that next-gen CPUs released around a year later won't be supported in those mobos"

 

This is the problem with Intel and this is what I really don't like.

Majority, because very very few people buy a motherboard with the intention to upgrade just the CPU within the next couple of years.  Extreme overclocker's who have to have the latest of everything every year and maybe those who upgrade every 2 years.  Almost everyone else keeps their systems for 4-5 years.

[Morgan MLGman]

Just now, mr moose said:

Majority, because very very few people buy a motherboard with the intention to upgrade just the CPU within the next couple of years.  Extreme overclocker's who have to have the latest of everything every year and maybe those who upgrade every 2 years.  Almost everyone else keeps their systems for 4-5 years.

I doubt that. Every concious customer would stop for a moment and think again about buying a brand-new platform that's EOL after a year. Which means no new motherboards after a year, no new features after a year, no new CPUs after a year and a very limited upgrade path (4C/8T).

[mr moose]

3 minutes ago, Morgan MLGman said:

I doubt that. Every concious customer would stop for a moment and think again about buying a brand-new platform that's EOL after a year. Which means no new motherboards after a year, no new features after a year, no new CPUs after a year and a very limited upgrade path (4C/8T).

Why would they need to buy a new motherboard?  The one they have will last 4-5 years.  

[Morgan MLGman]

4 minutes ago, mr moose said:

Why would they need to buy a new motherboard?  The one they have will last 4-5 years.  

Probably because their upgrade path is too limited and they don't want to buy a quad-core CPU in let's say 2020

[mr moose]

Just now, Morgan MLGman said:

Probably because their upgrade path is too limited and they don't want to buy a quad-core CPU in let's say 2020

if you buy a z270 and 7700k right now, it will be good for 3-4 years minimum (maybe even 5-6),  by then a new mobo and CPU is needed anyway.  Even if Intel make the 200 series compatible with coffee lake anyone buying now is still going to either miss out on being able to reuse their motherboard or will have to settle for what will be an EOL CPU for it anyway. 

[Morgan MLGman]

10 minutes ago, mr moose said:

if you buy a z270 and 7700k right now, it will be good for 3-4 years minimum (maybe even 5-6)

That is only a broad assumption, it entirely depends on the user and what he does with such CPU and you cannot guesstimate this as some people need the CPU horsepower which 7700K, while good at gaming, essentially lacks compared to the current X299/X399 or even R7 lineup. Even the R5 1600X has more 'raw' power under the hood due to the number of cores

 

Besides, this topic is about CPUs not being compatible with older, but same socket, motherboards which HEAVILY limits the upgrade path as you're only limited to quad cores... If you have an i5-6600K, an i7-7700K is not a significant upgrade, but if you could throw in an 8700K which will feature 6C/12T onto the same board, that's a different story when transitioning from 4C/4T (a.k.a. 3x fewer threads)...

[2FA]

5 hours ago, Morgan MLGman said:

Still, Kaby Lake to Skylake is what Devil's Canyon to Haswell was - no IPC improvements, just minor things including a new pointless chipset and higher average OCs.

People (including me) are mad because it's possible that it's Intel's artificial limitation that blocks previous LGA1151 chipsets from using Coffee Lake, created so people buy more motherboards. Why is the socket the same if previous chipsets don't fit? I've got no idea, but it was a retarded decision.

I get that, I'm not exactly happy either. I'm just aware of how Intel does things: 2011 vs 2011v3

[mr moose]

1 minute ago, Morgan MLGman said:

That is only a broad assumption, it entirely depends on the user and what he does with such CPU and you cannot guesstimate this as some people need the CPU horsepower which 7700K, while good at gaming, essentially lacks compared to the current X299/X399 or even R7 lineup. Even the R5 1600X has more 'raw' power under the hood due to the number of cores

 

Besides, this topic is about CPUs not being compatible with older, but same socket, motherboards which HEAVILY limits the upgrade path as you're only limited to quad cores... If you have an i5-6600K, an i7-7700K is not a significant upgrade, but if you could throw in an 8700K which will feature 6C/12T onto the same board, that's a different story when transitioning from 4C/4T (a.k.a. 3x fewer threads)...

 

It's not an assumption, there are many threads on here and reddit asking how often people upgrade.  There are one or two who upgrade every 2 years the rest is 4 years or more.    Genuinely very few upgrade often enough to make a change in socket support a big deal.

 

 

 

[MageTank]

13 hours ago, Drak3 said:

At the end of the day, it is what it is. Be it actual technical reason, or arbitrary lock, one will get a new board when they upgrade. And let's not all collectively kid ourselves here, if we have a 4770K+, we're not hurting that bad for an upgrade to our personal systems if getting a new mainboard is such a huge burden, if at all.

 

@MageTank I'd imagine the new socket comes down to either a DMI update that opens pins to other things, or an OC socket that has more than 1151 pins on Z series boards and K/X SKU processors. If you hear anything on it, PM me or or share it with the world, if appropriate?

 

Also, I don't see X299's next hexacore and the Coffee Lake hexacore interfering with each other too too much. I doubt CL will be quad channel and have more than 20 PCIe lanes (I can dream of PCIe goodness. Bite me.), whereas I can see quad channel and at least 28 lanes on the next X299 CPU. I see it more taking away from people wanting a Ryzen 5 or 7 that also check benchmarks first, at least a tad.

I was originally told that current sockets could potentially support it, and that pins that were currently inactive would be activated to support the additional cores, but this later changed to being an entirely new socket design that also improved memory speeds. It's still true that the chipsets themselves support coffeelake, but it's the socket pinout that changed.

 

Coffeelake will not be quad channel, but that's not the point. The point is, you have two 6-core offerings that both require a new platform compared to current Z170/Z270. If you are upgrading from say, Skylake or Kaby Lake, and can afford the cost/have the experience to cool X299, you are better off going that route as you will end up with a better upgrade path in the future, and have more memory bandwidth/capacity and potentially more PCIe lanes. CPU cost (assuming the 8700k is the same price as every other consumer i7 at launch) will only be about $60 difference between the two, and motherboard cost will be about $100-$150 difference (depending on quality of the board). If you already have 4 DIMM's of DDR4, you are basically good to go for quad channel. If you have two, you only need another 2 matching DIMM's for quad channel activation to take advantage of that boon. 

 

Just seems silly to "upgrade" to Coffeelake. For new builds, I can see it being viable depending on how high it clocks, and cost of entry (motherboard requirements, cooling requirements, etc). We are expecting very similar clock speeds to Kaby, so it wouldn't be unreasonable to see 5ghz Hexa-core CPU's on CFL. I just really wish Intel would unlock their entire product stack, so that people can buy i3's and i5's (non-K) and still get a respectable CPU out of it. A 5ghz+ 6 core i5 would absolutely smash Ryzen for the price, and offer consumers amazing price:performance without paying the "hyperthreading tax". 

[ravenshrike]

10 minutes ago, MageTank said:

We are expecting very similar clock speeds to Kaby, so it wouldn't be unreasonable to see 5ghz Hexa-core CPU's on CFL

Not without delidding. Toothpaste won't allow 6 cores to go above 4.8 even on a golden chip.

[Sauron]

They probably have to change the socket electrically to accomodate the higher core count cpus that they had no plan of releasing before the AMD truck hit them in the face.

On 8/2/2017 at 4:59 PM, mynameisjuan said:

AMD fanboy logic:

 

"Thats why I am going to ryzen, screw intel!!!"

 

*goes out to buy ryzen and needs to buy a new board also*

*doesn't have to wait for intel's next release*

*gets two more cores for the same price anyway*

*can overclock lower end chips*

(obvious joke is obvious but I think this needs to be added)

 

Not that I think you should upgrade your whole platform within a year anyway.

[Dylanc1500]

4 minutes ago, ravenshrike said:

Not without delidding. Toothpaste won't allow 6 cores to go above 4.8 even on a golden chip.

That toothpaste allows a 12 core chip to have an all core boost of 3.9 and a 18 core chip to have all core of 3.7. Just food for thought.

[PCGuy_5960]

27 minutes ago, ravenshrike said:

Not without delidding. Toothpaste won't allow 6 cores to go above 4.8 even on a golden chip.

Dow Corning TC-1996 is not toothpaste. Have you tried using it as a toothpaste? It tastes awful!

 

Seriously though, the thermal paste is fine, Intel's application method is what's causing problems

[ravenshrike]

14 minutes ago, Dylanc1500 said:

That toothpaste allows a 12 core chip to have an all core boost of 3.9 and a 18 core chip to have all core of 3.7. Just food for thought.

That's because the architecture puts out minimal heat at those levels. However heat generation increases exponentially past 4 Ghz. With 6 cores a golden chip is barely likely to get you to 4.8. Certainly not the 5 of the 4 core.

[MageTank]

51 minutes ago, ravenshrike said:

Not without delidding. Toothpaste won't allow 6 cores to go above 4.8 even on a golden chip.

That's a given on any i7 under my levels of stress. Find me a non-delidded i7 that can do Linpack MKL at stock clocks, lol. 

[Dylanc1500]

1 hour ago, ravenshrike said:

That's because the architecture puts out minimal heat at those levels. However heat generation increases exponentially past 4 Ghz. With 6 cores a golden chip is barely likely to get you to 4.8. Certainly not the 5 of the 4 core.

That's dependent on a lot of different factors a big one being voltage, if the voltages can stay low then that isn't necessarily true. They were able to increase clock/voltage efficiency on the 7740x. With further refinement they keep doing to their process, it is very plausible that they could hit 5ghz with two more cores. 

 

Its all speculation though.