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

Where do you expect Core i7-8700K's Turbo to land?

  • 3.8/3.9 GHz

    Votes: 0 0.0%
  • 4.0/4.1 GHz

    Votes: 3 23.1%
  • 4.2/4.3 GHz

    Votes: 6 46.2%
  • 4.4/4.5 GHz

    Votes: 3 23.1%
  • 4.6/4.7 GHz

    Votes: 1 7.7%

  • Total voters
    13
  • Poll closed .
The problem I always have with the "Z370 is a money grab" line of reasoning is that Intel can't possibly make a significant amount of money from motherboard sales to early adopters who already have Z270(or even Z170) and want to upgrade. As someone above mentioned, K-series CPUs(which is the CPU that anyone upgrading that frequently must be buying) are a small amount of their volume even compared to other desktop CPUs, let alone laptops. So what's the point of a half-assed money grab to a minimal percentage of customers? Doesn't make any business sense to me.
 
He indicates the Z270 chipset itself could be compatible, at least in theory, and thus Z270 boards could be made to support the new CPUs. This is not the same as existing Z270 based motherboards.
Even if they did make the Z270 chipset compatible, new motherboards would be required, as the CPU is using pins which were previously reserved (unused, left floating or all tied to ground). He is saying it would be pretty easy for them to route them up, and he's correct. It's still a new board required to do so.

Now whether or not the new CPUs really required a different pinout is a different question. I don't see anything which smells fishy at all though. Peak power will of course be higher even with the same TDP, which is not what many seem to think it is.

My take on that interview was that he thought other than a IME update and bios update that Z270 based Asus boards could handle CFL, and Intel was keeping it from happening.
 
14++ has a slight advantage. It's been pretty linear since SKL. Delidded SKL (14) hits 4.6-4.8Ghz. Delidded KBL (14+) hits 4.8-5.0 average and Delidded CFL (14++) hits 4.9-5.1 average. The same architecture, the same TDP, the same 14nm process...no reason why this couldn't be in the same motherboard if they chose. Even Asus engineers said the same thing in an article linked pages ago.

Asus engineers? Nope. The guy of the interview was a product manager, and the little technical details he provided were refuted in another article also linked pages ago.

Resume: The manager only said that Asus top KBL mobos "could" provide the power needed to support the known CFL chips. He didn't talk about the quality of the supplied power and he didn't talk about anything else. He doesn't know anything concrete about the role of the reserved pins in the new chipset, he only stated his belief those pins could be reserved for powering higher core-count future models (which means he cannot warrant older mobos to support future CPUs if released). He ignored that Intel had to consider the minimum common denominator among motherboards [1], and he ignored (dismissed) differences between old and new chipsets.

[1] "At one point in time, Coffee Lake processors may have been supported across some Z270 boards, but not all, with Intel deciding that the easier - and likely more profitable - option of limited board support to the latest chipset was the best course of action."
 
Last edited:
My take on that interview was that he thought other than a IME update and bios update that Z270 based Asus boards could handle CFL, and Intel was keeping it from happening.

Nope. Z270 could be compatible with CFL. But then it would be incompatible with SKL and KBL anyway.

What Asus just want is to try and sell you a 400$ board you struggle to hit good clocks on instead of a 150$ that does 5Ghz+ out of the box on easy mode.
 
And so it begins.

KthrGKh.png


Final Result
4 hours stable at 1.29v for 5 GHz. Peak temp 76C.

1.28v crashed after 30mins.
This DefineS + Noctuas are putting in work.

I think this is a 5.2 GHz chip with delid.

This is the guide I used.
 
Last edited:
I'm down to 1.28v I am starting to suspect RealBench is not very accurate.
I did play ACO for about 30mins at 1.29v.
 
I am going to add to this thread to those that are paying over $500 for 8700K's. Yes I owned one and yes it would do 5.2 ghz and yes the IPC was about 195-205 depending on how hard I pushed the chip. But for like $550, throw in some quad channel ram, OC the cache to 3000mhz and all cores to 4.7 @ 1.247 volts LLC 3 and you get this ....

Eat your heart you over priced, over hyped, over baked, and under delivered 8700K.... x299 is where it is at!

7820x_Single_Thread.png
 
Spending more for benchmark epeen?

No thanks :D

[I get what you're saying, and if I had an application that demanded more performance I'd even agree, but I don't, and I know that most enthusiasts fall into that category- unbeatable gaming performance on the 8700K is plenty, and it's plenty fast at everything else you throw at it too...]
 
Spending more for benchmark epeen?

No thanks :D

[I get what you're saying, and if I had an application that demanded more performance I'd even agree, but I don't, and I know that most enthusiasts fall into that category- unbeatable gaming performance on the 8700K is plenty, and it's plenty fast at everything else you throw at it too...]

Wasn't trying to argue or convince anyone not to get CLake. Just to consider if they have the budget that there might be a better buy in the SkylakeX chips for some folks given the current market price and availability.
 
Wasn't trying to argue or convince anyone not to get CLake. Just to consider if they have the budget that there might be a better buy in the SkylakeX chips for some folks given the current market price and availability.

You're spot on about market price- though Newegg, who actually has had continued availability for the last week or so, also ran a bit of a discount. Still around US$400 instead of the US$350 it should be at, but certainly attainable, and both the boards and RAM are cheaper.

And I certainly get the draw- the enthusiast in me would love to run a pair of 1080Ti's at full PCIe speeds, an NVMe RAID, a spinning RAID... but I've already spent enough to put together a blazing 8700k rig, and I just don't have the application needs to justify the big kids toys ;).
 
I am going to add to this thread to those that are paying over $500 for 8700K's. Yes I owned one and yes it would do 5.2 ghz and yes the IPC was about 195-205 depending on how hard I pushed the chip. But for like $550, throw in some quad channel ram, OC the cache to 3000mhz and all cores to 4.7 @ 1.247 volts LLC 3 and you get this ....

Eat your heart you over priced, over hyped, over baked, and under delivered 8700K.... x299 is where it is at!

7820x_Single_Thread.png

Using CB to try and show IPC/performance is silly because its just single cycle SSE at this point. And unless you OC the uncore quite a bit on SKL-X, then CFL-S runs in circles around it in uncore heavy loads like games tend to be. you can even run the uncore on CFL-S at 5Ghz or over.
 
Last edited:
8700K price right now on Newegg is $571.99 sold and shipped by Newegg! Why did it go up so much?
 
Screenshot (3).png

FuaJo.png
How bad is high voltage on an overclock if the temps are stable? I am getting 5Ghz using the Auto OC function in ASRock BIOS, AVX turned to 0, but it is using a lot of voltage. CoreTemp shows VID hitting up to 1.48v during games (it fluctuates), is this OK if my temps don't spike past 80c? Temps themselves average lower than my 7700k build, but that chip used less voltage to hit 5GHz (1.36).

FuaJo.png
 
Last edited:
View attachment 43565

FuaJo.png
How bad is high voltage on an overclock if the temps are stable? I am getting 5Ghz using the Auto OC function in ASRock BIOS, AVX turned to 0, but it is using a lot of voltage. CoreTemp shows VID hitting up to 1.48v during games (it fluctuates), is this OK if my temps don't spike past 80c? Temps themselves average lower than my 7700k build, but that chip used less voltage to hit 5GHz (1.36).

FuaJo.png
That is waaayaa too much. Your gonna kiss that chip goodbye in a year or less.

Manually set your offset to +0.015v over VID.

Set your load line to 3 or medium depending on your board.

An 8700k can hit 4.8ghz at 1.24 and 5.0 around 1.28 to 1.3... I have no idea why some people insist on cranking the voltage to the max limit and call it successful.

That auto voltage sounds like your load line is being set way to high. If your load voltages are that high its gonna cook that thing. But if idle voltage is that high and load voltage drops to less than 1.4 that is ok and safe.

Also your VID isnt 1.48 its your load line allowing the auto offset to go way to high over the actual baseline VID. For example I have a 7820x and I am all core x8 cores 4.8ghz at LLC 3 and 1.248 max load. An 8700k can easily OC better than skylake x.
 
Last edited:
Ok, thanks Tango. I went back to BIOS and set offset to 10 millivolts (15 not an option), LLC to "Level 3".

Now, Core Temp is still showing some higher VID spikes, but looking at CPU-Z shows the Core Voltage and it's not going any higher than 1.328, usually hangs around 1.248.

This means I'm not frying the chip, hopefully?
 
Last edited:
you'd know if that vid reporting is real or not via temps. lots of programs can't read new hardware correctly i suspect that is the case with core temp
 
Ok, thanks Tango. I went back to BIOS and set offset to 10 millivolts (15 not an option), LLC to "Level 3".

Now, Core Temp is still showing some higher VID spikes, but looking at CPU-Z shows the Core Voltage and it's not going any higher than 1.328, usually hangs around 1.248.

This means I'm not frying the chip, hopefully?
I recommend HwInfo 64. Even GamerzNexus swears by its accuracy compared to others.
 
I dropped down to 4.9 to see how low I can take the vcore. So far stable 2 hours Prime95 at 1.23V bios (hwinfo64 says 1.216V). Ran non AVX realbench for 1 hour.

Still working my way down. Next test should be sub 1.2V.
I cranked up llc to level 6, should flatline the vcore a bit.

Edit:

Passed 1hr of P95 at 1.184V, then finally got a 0x124 bsod in RealBench within 2 minutes. It looks like 1.200 or 1.216 is my vcore for 4.9 stable.

I won't be using prime95 again.
 
Last edited:
Hi guys. Just ordered the 8700k along asrock z370 mitx motherboard. Im planning to reuse my ddr4 memory from skylake build. The ram is https://m.newegg.com/products/N82E16820233970.

Will i be able to overclock with this setup? Thanks
Your ram is slow but that shouldn't hold you back from overclocking the processor. The motherboard may hold back your max a bit, but you can make some headway. Not sure exactly which model board you've got there but if it has solid power delivery and adequate vrm cooling that's pretty important.
 
Your ram is slow but that shouldn't hold you back from overclocking the processor. The motherboard may hold back your max a bit, but you can make some headway. Not sure exactly which model board you've got there but if it has solid power delivery and adequate vrm cooling that's pretty important.

Thanks for quick reply arnemetis. I will be using Fatal1ty Z370 Gaming-ITX/ac. Choices in mITX format is limited but newegg has this motherboard top rated that's why i bought it. Any experience with this board?

Does upgrading faster ram translates to better OCing? I will mainly game on this build.
 
Does upgrading faster ram translates to better OCing? I will mainly game on this build.
Ram can be OC separately, so you can OC the cpu no problem. Faster ram may gain a few percent better performance.

Most important is the cooler you plan on using as the 8700k gets hot
 
Ram can be OC separately, so you can OC the cpu no problem. Faster ram may gain a few percent better performance.

Most important is the cooler you plan on using as the 8700k gets hot

Good i cant afford faster ram right now. Few percent loss wont bother me a bit. I have a x61 kraken that i will be using to cool that 8700k.
 
I dropped down to 4.9 to see how low I can take the vcore. So far stable 2 hours Prime95 at 1.23V bios (hwinfo64 says 1.216V). Ran non AVX realbench for 1 hour.

Still working my way down. Next test should be sub 1.2V.
I cranked up llc to level 6, should flatline the vcore a bit.

Edit:

Passed 1hr of P95 at 1.184V, then finally got a 0x124 bsod in RealBench within 2 minutes. It looks like 1.200 or 1.216 is my vcore for 4.9 stable.

I won't be using prime95 again.

So how are your temps looking at 4.9 at those voltages? Your profile shows air cooling? I'm interested in exactly what you are trying to do there since I'll be using an air cooler also.
 
So how are your temps looking at 4.9 at those voltages? Your profile shows air cooling? I'm interested in exactly what you are trying to do there since I'll be using an air cooler also.
GPU heat was pushing the CPU up to 80C peak in AC:O, a bit high for my taste. This was at 90-100% GPU load and only 50% CPU. If my CPU load gets any higher it will go above 80C for sure. CPU stress testing by itself kept it around 70C-75C which is actually pretty cool, so it's not actually a heat problem with the CPU/OC.

A more efficient GPU, or an AIO on the CPU/GPU would solve it. Or I could turn my case fans up, but I don't want to make my PC any louder. Delidding is also an option down the road.

Right now under the same gaming load I'm in the high 60's at 1.216V 4.9 GHz.
 
Last edited:
  • Like
Reactions: Advil
like this
Hot damn. Maybe I should have waited for the 7820X to go on-sale like this.

I just picked one up for $550 and thought I was getting a deal. $499 at MC is a deal, but you pay sales tax ($30) and I'd have to drive to Philly to get it. If NE had them for $499, I would have jumped all over it.
 
Back
Top