ASRock: Coffee Lake CPUs not compatible with 200 series motherboards

[happy medium]

My last AMD motherboard was a GA-MA785GM-US2H and it supported 2 , 3, 4, and 6 core cpu's.
45nm, 65nm and 90nm @ 35 to 140 watts.

http://www.gigabyte.us/Ajax/SupportFunction/Getcpulist?Type=Product&Value=3447

That was 10? years ago.

I don't see any reason but greed that intel couldn't
make a motherboard chipset that lasts more than a year or 2.

 

[happy medium]

"In previous leaks, we confirmed that their would be a new socket known as LGA 1151 V2 which is pretty much the same thing but has a slightly different electrical pin configuration. This means that current LGA 1151 socket boards are not able to support Coffee Lake chips and the same would be true for 6th and 7th generation CPU support on the upcoming 300-series platform which is built entirely for Coffee Lake processors."

http://wccftech.com/intel-coffee-lake-cpu-not-compatible-200-series-motherboards/

 

[VirtualLarry]

Huh, so Intel's making one-hit-wonders now? (CPU / chipset generations) Talk about an 80s throwback! Bring back the dancing bunny suits! (Ok, so that was the 90s, not the 80s, but go with it.)

 

[moinmoin]

Intel stopped doing tick-tocks for processors but kept it for the chipsets. Worst of both worlds.

 

[Vaporizer]

Well it seems that they are currently focussing on customers. Less artificial segregation, no artifical cutting on HEDT chips( e.g PCI lanes), good thermal interface to enable good oc, suitable upgrade path on platform (tick/tock)....
Oh wait!

 

[beginner99]

Repeating myself from other thread but this was obvious. Why? 6-core CFL will use a lot more power than a 7700k did. Don't get fooled by the tiny TDP difference (91w to 95w). We have seen with Skylake-X that intel is doing more and more TDP shenanigans, eg tdp same as broadwell-e but power use much higher. I bet same will happen with CFL 6-core, especially when OCing to say all-core 4.7 ghz or something. z270 socket would probably go up in flames with that power use.

 

[PingSpike]

If you bought a z270 board hoping for a drop in upgrade you got screwed. But honestly...you should have known better than to do that.

I agree with DaveSimmons in that the biggest bummer, besides having to pay more for newer boards is that you're going to have to deal with platform teething issues.

 

[Topweasel]

Kind of expected a new pin count. But I have been saying this for a long while. Intel's consumer platforms are only good for the time period they worked out for OEM's which is 2 years. Kind of surprising that 300 series isn't going to be backwards compatible with SL/KL all things considered. But there was no way Intel was going to lax on that 2 year window no matter what kind of consumer popularity they could gain by not forcing an unneeded upgrade.

 

[Malogeek]

You should ask yourself if you really need that 6 core coffeelake if you've got a 6700k. It's not a CPU requiring upgrade. Intel probably knows the number of people doing CPU upgrades is negligible so it doesn't care about them.

Before Ryzen the idea of an extra 2 cores and 4 threads would have been huge to an enthusiast community for a lower end (not HEDT) mainstream gaming CPU. Oh how perception has changed in such a short time.

 

[formulav8]

Don't get fooled by the tiny TDP difference (91w to 95w)

Yeah I've been wondering about CFL 6 core only 4w higher tdp with the same, although tweaked process. Although tdp isn't technically the power consumption it has a relation to it and would like to see the power numbers against 7700k.

Although @ 6 cores it runs 200 mhz slower than 7700k. So that may explain some.

 

[scannall]

If you bought a z270 board hoping for a drop in upgrade you got screwed. But honestly...you should have known better than to do that.

I agree with DaveSimmons in that the biggest bummer, besides having to pay more for newer boards is that you're going to have to deal with platform teething issues.

That, and it will be a 1 generation board. Next generation will be a different socket.

 

[dullard]

Yeah I've been wondering about CFL 6 core only 4w higher tdp with the same, although tweaked process. Although tdp isn't technically the power consumption it has a relation to it and would like to see the power numbers against 7700k.

Although @ 6 cores it runs 200 mhz slower than 7700k. So that may explain some.

Running slower lowers the power, but also the Coffee Lake 14nm++ transistors themselves use less power at the same performance level:
https://en.wikichip.org/wiki/File:intel_14nm++.png

The question that I have that is not yet answered though is how much time the Coffee Lake processors will run at full turbo. Kaby Lake had enough spare TDP that for most people here (with good cooling) you could assume that it usually would run in turbo mode and rarely ran at base frequencies. What I would like to know is if Coffee Lake throttles back to base more often than Kaby Lake.

 

[kwalkingcraze]

LGA1151 V2 looks identical to FM2+ we'll be seeing. Backward compatible to LGA1151 like FM2, but not forward compatible to LGA1151 V2 and FM2+.

 

[TemjinGold]

Wwwhhaaat? I distinctly remember some people were 99% certain it would work!

 

[crashtech]

Wwwhhaaat? I distinctly remember some people were 99% certain it would work!

Yeah, where are they now, lol.

 

[formulav8]

14nm++ transistors themselves use less power at the same performance level

If that much improvement then Intel did quite well on their 3rd version.

 

[moinmoin]

I thought the need for new boards was common knowledge as even Wikipedia mentioned CFL to be using a 300 series chipset. Looking in its history that info was inserted and sourced in late March/early April.

 

[Abwx]

If that much improvement then Intel did quite well on their 3rd version.

Or that their first 14nm was mediocre to begin with, at the time i stated that the perf/watt improvement over 22nm was a meager 13-15% and close to nothing if we account for Haswell IVR, indeed this 14nm++ provide about the benefit of a real full node shrink in respect of....22nm.

 

[Jan Olšan]

Ironically, in the 754, 939 and AM2 era, Intel was seeing as the best in this because they keep the 775 socket and you could slap a old cpu intro a new mb.

They were not compatible, 754 and 939 users got screwed, badly, most AM2 boards could be never upgraded to place a AM2+ cpu on them, yeah, you could place a AM2 cpu intro AM2+ board, yeah, FM1 lasted a year and replaced by FM2... some FM2 boards could not be upgraded to accept FM2+ cpus, depended on OEM, with AM3? the exact same thing, this time im going to take the risk and say that almost no AM3 MB could be upgraded to accept AM3+ cpus. AM1 had no upgrade at all.

And Intel did the same, with 775 chipsets, 1155 and 1156... etc.

As a hardware collector you should know this.

Who is willing to ensure that AMD will not launch a "AM4+"? tecnically is not changing the socket, but compatibility depends on OEM. They did this several times now and got burned twice, so im not placing my hands on the fire for them this time.

1) Socket 754 was indeed not very good, but at least you could put both 130nm and 90nm CPUs in it. its purpose was to make boards cheaper, so while its production might have been a long-term mistake, it was not just soem ill will.

2) Socket 939 was abandoned due to switch to DDR2, so there is no grounds for criticism at all. And moreover, it supported 130nm, 90nm chips AND dualcores!

3) Which brings up to the LGA 775. As was already said, there were numerous compatibility breaks. When Core 2 came, and then when 45nm Core 2 came (Penryn) - those both cut compatibility with older boards like Coffee Lake now, or like Broadwell desktop chips. But that was not everything.

775 boards made before introduction of dualcore Pentium Ds could not be upgraded with them and you had to buy new boards too. ANd I think there was one similar schenanigan when Intel raised current requirements - again, newer chips unsupported in previous motherboards. There might have been something with 1333MHz FSB but I am not sure.

I hope you see why I felt the need to defend socket 939. Sure, it was unpleasant for me that the dualcores for it never got cheap while the AM2 ones did. But that is a different problem - the platform's life was not cut short artificially, it ended due to migration to DDR2, which was already cheaper than DDR1 in 2006 and thus needed to be adopted.

 

[jihe]

Now that is what you call a 'dick move'

 

[Shivansps]

1) Socket 754 was indeed not very good, but at least you could put both 130nm and 90nm CPUs in it. its purpose was to make boards cheaper, so while its production might have been a long-term mistake, it was not just soem ill will.

2) Socket 939 was abandoned due to switch to DDR2, so there is no grounds for criticism at all. And moreover, it supported 130nm, 90nm chips AND dualcores!

3) Which brings up to the LGA 775. As was already said, there were numerous compatibility breaks. When Core 2 came, and then when 45nm Core 2 came (Penryn) - those both cut compatibility with older boards like Coffee Lake now, or like Broadwell desktop chips. But that was not everything.

775 boards made before introduction of dualcore Pentium Ds could not be upgraded with them and you had to buy new boards too. ANd I think there was one similar schenanigan when Intel raised current requirements - again, newer chips unsupported in previous motherboards. There might have been something with 1333MHz FSB but I am not sure.

I hope you see why I felt the need to defend socket 939. Sure, it was unpleasant for me that the dualcores for it never got cheap while the AM2 ones did. But that is a different problem - the platform's life was not cut short artificially, it ended due to migration to DDR2, which was already cheaper than DDR1 in 2006 and thus needed to be adopted.

You are right about 775, it had backward support, but forward support was indeed a mess, yet in those years AMD was seeing as the one who changed sockets too much, as 775 remained static and supported from DDR1 to DDR3. It was a technical reason i know.
I remember i started with 754 and them switched to AM2(i had a Xp1700+ before this), while i agree with you that the 939 to AM2 switch was because of DDR2, people was not happy at all back them. 939 was short lived, and as you said, dual cores were expensive. 939 should have never existed at all,
Personally im not happy at all with AM2 either, neither of my two AM2 boards recibed bios update for AM2+ cpus (Asus m2n4 sli and MSI K9AGM2), and AM2 cpus had bad TIM problems, TIM dried after 1+ year use on my X2 4200+, and many, many people had this issue as well.
Then i switched to AM3 (Gigabyte GA-MA770T-UD3) and Phenom 2 X3 710, nice cpu, but i got screwed again with AM3+, switched to Intel the exact same day FX8150 came out(brought a 2500K) and lasted for 6+ years of constant abuse, OC, and near top of the line performance.

So you can see why i dont trust AMD too much and i find funny to read that AMD is better at keeping motherboard compatibility. And thats not considering what they did with FM1 and FM2, that was 754 and AM2 all over again.

I decided to give AMD a second chance with Ryzen, i have a R7 1700 now, but i really dont trust anything they said, im fully expecting them to release AM4+ with back support for wharever reason and forward support depending on oem.

 

[kirbyrj]

The kicker is Z370 isn't even a new chipset. It's just a rebadged Z270. I'm not even sure that the whole v2 distinction for "extra power" is real. Seems like enough ES chips were in the wild running on Z270 boards.

 

[Insert_Nickname]

You are right about 775, it had backward support, but forward support was indeed a mess, yet in those years AMD was seeing as the one who changed sockets too much, as 775 remained static and supported from DDR1 to DDR3. It was a technical reason i know.
I remember i started with 754 and them switched to AM2(i had a Xp1700+ before this), while i agree with you that the 939 to AM2 switch was because of DDR2, people was not happy at all back them. 939 was short lived, and as you said, dual cores were expensive. 939 should have never existed at all,

The reason 775 supported DDR(1)-trough-DDR3 was the memory controller was in the northbridge rather then CPU. It connected via FSB to the CPU, so it was rather "easy" to support newer memory standards on 775. You just needed a new chipset with a new memory controller. AMD sockets newer then Socket 462 had the memory controller integrated into the CPU. A disadvantage in compatibility terms, but a very potent advantage in performance. Result was you got Socket 754 (64-bit single channel DDR), Socket 939 (128-bit dual channel DDR) and AM2 (128-bit dual channel DDR2).

I disagree it was 939 that should not have existed, rather it was 754. But I can see why they did it for cost reasons. You're not mixing the whole Socket 940 debacle with Socket 939?

As a side note, because of 775 having the memory controller in the northbridge, a rare few motherboards can run Pentium4/Ds with DDR3 memory. A little interesting quirk...

and AM2 cpus had bad TIM problems, TIM dried after 1+ year use on my X2 4200+, and many, many people had this issue as well.

Yup. Those early dual cores ran hot. A little Arctic Silver 5 or similar fixed that quickly.

So you can see why i dont trust AMD too much and i find funny to read that AMD is better at keeping motherboard compatibility. And thats not considering what they did with FM1 and FM2, that was 754 and AM2 all over again.

Both AMD and Intel are equally bad in that respect, except for the corner cases.

 

[Sonikku]

This was an ill timed move on Intel's part when AMD's value is kicking them in the ass. AM4 is going to last for several years, reducing the long term cost to using AMD as a platform.

 

[Shmee]

I wonder if there will be a BIOS/pin mod to fix this