Intel's 8th Generation Core Family - Coffee Lake (LGA 1151, 6C/12T)

[Dayman]

https://videocardz.com/70978/intel-preparing-multiple-6-core-coffee-lake-cpus (Article)
https://www.cpchardware.com/coffee-lake-approche/ (Original source in french)

Looks like we got base clocks/cache and core configs (I7/i5)
I7s:

i7-8700K(6t/12t) 3.7 GHz base frequency 95W TDP 12mb L3

i7-87006t/12t) 3.2 GHz base frequency 95W TDP 12mb L3

i5s:

i5-8600K(6c/6t) 3.6 GHz base frequency 95w TDP 9mb L3

i5-8400(6c/6t) 2.8 GHz base frequency 65w TDP 9mb L3

Coffee Lake Mobile:
High-end 6-core versions, currently being validated at 2GHz and 45w TDP

Coffee Lake U range will offer 4 cores with HT, and a GT3e graphics chip (with eDRAM), all in a thermal envelope of 28W.

One last think, I was browsing CanardPC twitter after this and they claim 200series and some 100series boards will support CFL. Whether or not this is true is to be seen.

 

[VanGoghComplex]

https://videocardz.com/70978/intel-preparing-multiple-6-core-coffee-lake-cpus (Article)
https://www.cpchardware.com/coffee-lake-approche/ (Original source in french)

Looks like we got base clocks/cache and core configs (I7/i5)
I7s:

i7-8700K(6t/12t) 3.7 GHz base frequency 95W TDP 12mb L3

i7-87006t/12t) 3.2 GHz base frequency 95W TDP 12mb L3

i5s:

i5-8600K(6c/6t) 3.6 GHz base frequency 95w TDP 9mb L3

i5-8400(6c/6t) 2.8 GHz base frequency 65w TDP 9mb L3

Coffee Lake Mobile:
High-end 6-core versions, currently being validated at 2GHz and 45w TDP

Coffee Lake U range will offer 4 cores with HT, and a GT3e graphics chip (with eDRAM), all in a thermal envelope of 28W.

One last think, I was browsing CanardPC twitter after this and they claim 200series and some 100series boards will support CFL. Whether or not this is true is to be seen.

Oh boy oh boy oh boy

 

[-Sweeper_]

The Turbo clocks though...

200-series / 100-series support is new to me. Up until late June they were set on restricting to Z370 and future chipsets. Change of plans?

 

[Shintai]

The 7700 is 65W and 3.6Ghz base, 4.2Ghz turbo. I doubt adding 2 more cores and using 14nm++ would result in a 3.2Ghz base clock while increasing the TDP ~50%

Same for mobile, already 3.1Ghz base on a quad at 45W. 2Ghz with 6 cores would be meaningless if it was true.

I say something is wrong in the CPC rumour

 

[-Sweeper_]

I believe they confused it Shintai. Should be 65W for 8700 non-K. Mobile either bogus or an early ES. Remmember this 4+3e CFL-U SKU is expected only in late Q1-2018.

 

[juanrga]

There's 0% science in your post.
I think what you're saying is
"It's 14nm++ so I'd really like it to run the same speeds as the 7800X, minimum"

Just before your post I explained some of the science behind the prediction that the clocks would be higher... and first leak confirms clocks are higher.

7800X 6C 3.5 GHz
8700K 6C 3.7 GHz

 

[juanrga]

i3 --> 4C + HT
i5 --> 6C
i7 --> 6C + HT

would be fantastic.

Confirmed that the i5 8600K and 8400 are six-cores without hypertreading

 

[juanrga]

The Turbo clocks though...

The base clock is 200MHz higher than the 7800X. The 4.5GHz or higher turbo is warranted.

 

[Shintai]

The Turbo clocks though...

200-series / 100-series support is new to me. Up until late June they were set on restricting to Z370 and future chipsets. Change of plans?

Oh ye the turbo clocks are nice

And then we got it confirmed, CFL on all platforms. CFL being IMVP8 and IMVP9 compliant. Only ICL and TGL requires IMVP9 that you see on 300+ series mobos.

 

[-Sweeper_]

Turbo looks great, it's better than 7700K (4.7 GHz)

 

[Shintai]

4.7Ghz is for 1 core. Then people can guess the 2, 4 and 6 core turbo

 

[EuphoricRage470]

4.7Ghz is for 1 core. Then people can guess the 2, 4 and 6 core turbo

Damn. Almost regretting going x299...that is awesome!

 

[FrozenSun]

i3 --> 4C + HT
i5 --> 6C
i7 --> 6C + HT

would be fantastic.

I'd more expect the i3 to just be a flat 4 core. I feel like 4C + HT vs 6 core would be too close in performance. I suppose they could simply use clock speed as the distinction instead.

 

[Skyblue]

Well, the I9s have some more cache. Lets see the real world results..

 

[Deleted whining member 223597]

You know what sucks, that Intel will probably use their cheap thermal garbage again instead of the 2 cents it costs to solder it. Hopefully the temps are more like skylake than Kaby lake.

 

[STR]

You know what sucks, that Intel will probably use their cheap thermal garbage again instead of the 2 cents it costs to solder it. Hopefully the temps are more like skylake than Kaby lake.

They are *highly* unlikely to be able to solder Coffee Lake dies. Too small. Too thin. Having said that, they might not use the same TIM, or use a better process to eliminate their quality control issues with the IHS gap.

 

[Deleted whining member 223597]

They are *highly* unlikely to be able to solder Coffee Lake dies. Too small. Too thin. Having said that, they might not use the same TIM, or use a better process to eliminate their quality control issues with the IHS gap.

I sure hope so, I mean people have been complaining about it since Ivy Bridge and there’s obviously a solution(use liquid metal) that works.

 

[Immolation1991]

jesus, and I just picked up a used 6700k, because I thought 1151 is not supported.

 

[STR]

I sure hope so, I mean people have been complaining about it since Ivy Bridge and there’s obviously a solution(use liquid metal) that works.

Is there any proof that liquid metal TIMs will last the 5+ years Intel requires of its CPUs? Heat/cool cycles on the order of 60c/second can cause a lot of mechanical stress if the material isn't flexible enough to handle it.

 

[Shintai]

Intel isn't using solder anymore in any product. Xeon Phi, Xeon series, I series etc. If you still nag over it its because you dont understand the reason.

 

[Deleted whining member 223597]

Is there any proof that liquid metal TIMs will last the 5+ years Intel requires of its CPUs? Heat/cool cycles on the order of 60c/second can cause a lot of mechanical stress if the material isn't flexible enough to handle it.

I’m not sure, my 3770K isn’t delidded. But I haven’t read about any failures. I just don’t get how they randomly seem to do Ok like with skylake and then terribly with kaby lake. Maybe the higher clock speeds.

 

[aztekk]

I consider the usage of TIM between the die and IHS an advantage. After delidding you can run it bare die to achieve superior cooling than having a chunk of metal between the heatsink and the die. It's your choice, unlike with a soldered IHS which is much more difficult/impossible to delid in a home environment. Last time I checked it made a huge difference in thermal management, I would go so far as to sugest if we didn't have IHS's on CPU's from the factory then we wouldn't have had watercooling transition from being a niche, extreme cooling method to a typical mainstream cooling solution.

I suppose it does help with longevity of the CPU if it's stored improperly (I broke many chips in the Athlon XP days this way lol) so having the IHS and making it somewhat easily removable (TIM) honestly seems like a good compromise.

 

[Deleted whining member 223597]

Intel isn't using solder anymore in any product. Xeon Phi, Xeon series, I series etc. If you still nag over it its because you dont understand the reason.

I couldn’t care less over some reason that you think matters, the temperatures matter to the consumer. If you wanna make excuses for Intel, go ahead, but I won’t. Even if I choose Intel over AMD, I’m not going to blindly accept all of their decisions.

 

[TaintedSquirrel]

I consider the usage of TIM between the die and IHS an advantage. After delidding you can run it bare die to achieve superior cooling than having a chunk of metal between the heatsink and the die. It's your choice, unlike with a soldered IHS which is much more difficult/impossible to delid in a home environment. Last time I checked it made a huge difference in thermal management, I would go so far as to sugest if we didn't have IHS's on CPU's from the factory then we wouldn't have had watercooling transition from being a niche, extreme cooling method to a typical mainstream cooling solution.

I suppose it does help with longevity of the CPU if it's stored improperly (I broke many chips in the Athlon XP days this way lol) so having the IHS and making it somewhat easily removable (TIM) honestly seems like a good compromise.

I would agree with you if it was possible to easily remove the IHS with your bare hands (no adhesive). Like removing any other kind of heatsink. Maybe if Intel used screws on the edge of the IHS to hold it down.

As it stands now, it's risky and you have to buy a kit to do it, which means like 99% of people won't bother.

 

[Dayman]

Intel isn't using solder anymore in any product. Xeon Phi, Xeon series, I series etc. If you still nag over it its because you dont understand the reason.

I've actually being trying to find out why they don't use it. Why don't they?

 

[AbRASiON]

Those frequencies seem too high, considering 7700k heat and TDP and the 140W of the x299 6 core.

 

[juanrga]

Those frequencies seem too high, considering 7700k heat and TDP and the 140W of the x299 6 core.

The 7700k uses a less optimized process, and the 140W of the SKL 6 core are including power hungry AVX512 units lacking in Coffee Lake.

 

[Shintai]

I've actually being trying to find out why they don't use it. Why don't they?

Because a too big amount of chips dies due to thermal cycling. I could easily fill a bag with (semi) dead Xeons due to thermal cycles over the years. And for stock operation the benefit is pretty much 0. How many forum posts haven't we seen either with the same issue where always good chip now suddenly runs extremely hot and throttles. And if you want to go crazy on the OC, delid it.

 

[-Sweeper_]

Hint: The entry level Core i5 from the new lineup has enough power to match/beat 7700K in MT apps, that's how good Turbo is for this 14nm++ lineup. Obs: I'm assuming it can sustain the rated Turbo at 65W.

 

[Brahmzy]

Now this sounds like a worthy upgrade. I think I heard my wallet whimpering...

 

[RPGWiZaRD]

Now this sounds like a worthy upgrade. I think I heard my wallet whimpering...

At least it won't be that expensive, I think these CPUs (8700K and 8600K) will be the new bang-for-buck alternative, I'm personally expecting $359 for 8700K compared to $389 Core i9 7800X 6C/12T, sounds reasonable. I'm also expecting for example in less well threaded apps/gaming this CPU will even be up there with Intels 8 core Skylake X $599 and even beat AMD Threadripper 16 core $999 in quite a few cases and make Ryzen X1800 8 core look way way overpriced (I'm expecting $80'ish price drop soonish) as it should lose handily if you take like 20 tests and summarize average result (beating in like 15 cases out of 20). Add the overclocking potential with a high clock frequency and superb IPC for this optimized 14nm++ process and I think the bang-for-buck king will be warranted. But time will tell but I've got a good feeling about this.

I'm not expecting a huge IPC increase vs Kaby Lake though as Intel would hurt their Skylake X/Kaby Lake X sales too much, with that in mind I think it's reasonable to not expect more than about 3% worst case, 12% best case 5~7% avg IPC increase vs Kaby Lake tops but it's good to increase it a bit with incoming Threadripper so they can still play that superb IPC + clock frequency advantage card, aka. the mainstream PC gamer's alternative. Any IPC + clock frequency scaling/leakage improvement they can take advantage of is still the best medicine to fight AMD which rely on an architecture that sacrifices clock speeds and IPC potential for core count so software have to be properly multithreaded for you to get your money's worth, otherwise you lose some instead.

 

[Armenius]

Because a too big amount of chips dies due to thermal cycling. I could easily fill a bag with (semi) dead Xeons due to thermal cycles over the years. And for stock operation the benefit is pretty much 0. How many forum posts haven't we seen either with the same issue where always good chip now suddenly runs extremely hot and throttles. And if you want to go crazy on the OC, delid it.

This, and the problem was more to due with the gap between the die and IHS, not that they used TIM instead of solder. There wouldn't be such a large delta between the two if the IHS made good contact with the die to begin with.

 

[Gideon]

Someone favorite site is showing the regression of performance on these new chips. 7800X is slower then then the 7700K it replaced, thats pretty sad.

 

[Shintai]

7800X didn't replace the 7700K. The 8700K replaces the 7700K.

 

[Gideon]

7800X didn't replace the 7700K. The 8700K replaces the 7700K.

Got confused on their naming scheme. Either way the 7800X is a let down in gaming. Forgot for a moment looking at the benchmarks that it was a 6 core instead of a 4 core.

 

[Shintai]

Got confused on their naming scheme. Either way the 7800X is a let down in gaming. Forgot for a moment looking at the benchmarks that it was a 6 core instead of a 4 core.

Of course it is if you use stock.

Uncore is clocked 900Mhz lower, base clock is 700Mhz lower and turbo clock is 500Mhz lower.

8700K will be the new gaming king.

 

[Immolation1991]

won't they come in August, or what is the expected release? The article linked earlier says "in few months".

 

[Raendor]

With alleged 100/200 chipset support that certainly looks intriguing. And if it's basically the same Skylake/Kabylake with added cores - I don't see a reason for this support not to exist, especially if Intel wants to be seem more consumer-friendly with AMD stepping on it's toes right now. Definitely will pick some hexacore if that's the case. Will be last upgrade before full platform revamp.

 

[Gideon]

Of course it is if you use stock.

Uncore is clocked 900Mhz lower, base clock is 700Mhz lower and turbo clock is 500Mhz lower.

8700K will be the new gaming king.

They overclocked the 7800X 4.7 and it still lost to a stock 7700K. The 7800X is just a bust on the gaming front.

 

[juanrga]

They overclocked the 7800X 4.7 and it still lost to a stock 7700K. The 7800X is just a bust on the gaming front.

Didn't I say you just above that OC the mesh improves gaming performance a lot of?

Up to 25% higher performance with a mesh OC

I find amazing that many reviews tested OC memory on RyZen to reduce latency and improve performance in games compared ot a stock comparison, but it seems no review of SKL-X has tested gaming performance with OC mesh to reduce latency.