Z170 CLAIMED to be hacked to support Coffeelake

[corpsejockey]

I've never even mentioned overclocking. I wouldn't expect much if any headroom at all. Increasing vcore is definitely going to bump up wattage no question, but stock 8700 hits around 160w full load. I submit to the idea that z270 power delivery was inadequate per intel specs for cl. I remember reading somewhere that if z270 had 1 more vcc pin on the socket, it would have fallen in spec for cl. I still think the mod is interesting and viable.
Damn wingman, you smart. I like you.
What about the other angle? Had the chip been slightly more efficient, would cl + z270 been possible? Again, not busting Intel's chops here. They produced a chip with 50% more cores @ 37% more power consumption. Just seems with a little more time and effort it could have happened.

 

[wingman99]

This is the reason Intel went with a new Z370 motherboard.

There were three key features of the Z370 platform that had to be made (according to Intel) which kind of validated their claim for the electrical pin routing change on the same LGA 1151 socket that had been released and available since Skylake back in 2015. Intel states that they improved power delivery for 6 core processors, enhanced package power delivery for overclocking and enhanced a series of memory routing features to support DDR4-2666 MHz.
https://wccftech.com/intel-coffee-lake-100-series-motherboard-support/

When compared, the Coffee Lake processors have 391 VSS (Ground) pins which is an increase of 14 compared to Kaby Lake, 146 VCC (Electrical) pins which is an increase of 18 pins compared to Kaby Lake and about 25 pins that are reserved and a decrease of 21 pins from the 46 reserved on Kaby Lake.
https://wccftech.com/review/intel-core-i7-8700k-core-i5-8600k-core-i5-8400-cpu-review/

 

[corpsejockey]

Yo, wingman... I know why intel says they did what they did. It still doesn't sit quite right with me, but again... Intel's official spec states it wasn't possible so be it. Perhaps it's because a z170/270 board build to absolute minimum spec is truly not capable? I don't have a clue about really budget boards or lower end chipsets. Haven't had one and never cared to research it. However, most midrange z series mobos have more than adequate vrm delivery. Most mid to upper end boards with 4 phase and doublers (not even true 8 phase) can deliver in the neighborhood of 200amps (max) to the socket... Yes 200amps. Again, this is more than capable of feeding a stock 8700 at peak draws. It even leaves a little headroom for oc. I'm not concerned with killing my vrm or melting the board. Someone looking at this thread and thinking they can take any old board and attempt the mod without doing any research on what they are working with in terms of hardware should not do it. If you want some intel ranting I would be more than happy... They've done things worse with cl than no backwards compatibility lol

 

[Alaric]

enhanced a series of memory routing features to support DDR4-2666 MHz.

Pretty sure Skylake could handle DDR4-2666 MHz. In fact, I know at least one guy who is running his DDR4 memory @3733 MHz. LOL. I'm going to oversimplify here, I'm sure, but if AMD can have a socket that came out for the Phenom and Athlon 64 chips and it supports a FX 9590, Intel should have been able to handle the bump to 6 cores on the Skylake socket.

 

[wingman99]

Pretty sure Skylake could handle DDR4-2666 MHz. In fact, I know at least one guy who is running his DDR4 memory @3733 MHz. LOL. I'm going to oversimplify here, I'm sure, but if AMD can have a socket that came out for the Phenom and Athlon 64 chips and it supports a FX 9590, Intel should have been able to handle the bump to 6 cores on the Skylake socket.

It's good to know you have more knowledge and experience than Intel's fleet of micro electronic engineers. What what would you do to run and overclock the the coffee lake 6 core on any Z270 with out the board burning up or overloading? Also what would you do to guarantee every Z270 around the world will do 2666 MHz memory speed without any problem at all?

 

[Alaric]

what would you do to guarantee every Z270 around the world will do 2666 MHz memory speed without any problem at all?

For that, build it on a Z170 chipset instead?

It's good to know you have more knowledge and experience than Intel's fleet of micro electronic engineers

Don't recall claiming that. I think I said

I'm going to oversimplify here

And honestly, if you think the engineers have a say equal to the accountants and marketing, I would disagree. Two of my uncles were engineers, and good at it. And they would rather have invented Star Trek's transporter than "kludge" new spark plugs in a car. They wanted a clean sheet of paper for everything, even if the previous application just needed a screw. A proper design from scratch is admirable. It's also easier when the customer is paying for it.

I don't claim to be an engineer, but Intel essentially had the market to themselves for a decade or more, and could (and did) do as they pleased, because there weren't any real options (according to Intel's marketing dept., anyway). Their collective actions after Ryzen seem to indicate a rudderless ship. Again, that's how it appears from the outside. And Intel's corporate history in the CPU arena is rife with examples of the marketing department making up revenue the engineers couldn't pull off. I acknowledge I know far less about the details than Intel. That doesn't mean I'm going to blindly follow Big Blue off a personal finance cliff because "the engineers said so".

 

[EarthDog]

What what would you do to run and overclock the the coffee lake 6 core on any Z270 with out the board burning up or overloading?

raise vcore and multiplier...like any other.

Also what would you do to guarantee every Z270 around the world will do 2666 MHz memory speed without any problem at all?

I wouldnt do anything.

All they did was up the JEDEC spec.

 

[Alaric]

raise vcore and multiplier...like any other.

I just looked at the performance numbers for the 8700k. Yes, it has two more cores than mine-that run at 3.7 GHz. I think Turbo only raises one or two cores, for a short period of time, correct? I'm holding four cores at 4.7 GHz 24/7 at the moment, with HT turned on. And my chip is far from golden. There are numerous examples of better binned, better cooled Skylake i7s running 4 cores at 4.8+ GHz on Z170 chipsets. The "moar volts" argument just doesn't hold up to existing evidence to the contrary. There may have been other, better reasons, but the excuse they gave is a poor one.

 

[wingman99]

Yo, wingman... I know why intel says they did what they did. It still doesn't sit quite right with me, but again... Intel's official spec states it wasn't possible so be it. Perhaps it's because a z170/270 board build to absolute minimum spec is truly not capable? I don't have a clue about really budget boards or lower end chipsets. Haven't had one and never cared to research it. However, most midrange z series mobos have more than adequate vrm delivery. Most mid to upper end boards with 4 phase and doublers (not even true 8 phase) can deliver in the neighborhood of 200amps (max) to the socket... Yes 200amps. Again, this is more than capable of feeding a stock 8700 at peak draws. It even leaves a little headroom for oc. I'm not concerned with killing my vrm or melting the board. Someone looking at this thread and thinking they can take any old board and attempt the mod without doing any research on what they are working with in terms of hardware should not do it. If you want some intel ranting I would be more than happy... They've done things worse with cl than no backwards compatibility lol

I have only used the cheapest motherboards for overclocking with Intel since 2001 because there power specification have been solid until coffee lake. With Coffee lake overclocking there is a lot of trouble with adequate power limit for the CPU, VRM overheating then throttling, current limit throttling. I'm helping people all the time with current limit throttling with overclocking on high-end motherboards. Intel should of gone with a lager socket like AMD ryzen also HEDT for more current and ground pins on coffee lake, maybe they will on Z390.

I would rather see a new motherboard than Intel limit the overclocking like they have done with high loads causing current limit throttling that has to be overridden on the Z370. I know what your saying about upgrade path, however AMD has not had solid motherboards in the past for overclocking there range of processor upgrades, that is why I'm with Intel. If Intel did not sell overclocking processor and motherboards there would be no discussion on this subject.

I can careless about any Z boards VRM's, they all have to run within Intel's specifications of maximum AMPs and that is well above what I see most people use with overclocking coffee lake. I care about the board and pins and pad on the bottom of the CPU burning up do to the increased current with overclocking.

Have you ever seen the LGA pins, pad, motherboard PCB burn?

 

[wingman99]

raise vcore and multiplier...like any other.
I wouldnt do anything.

All they did was up the JEDEC spec.

That does not read like good engineering step to prevent the pins tracers from burning up do to the increased amps of 6 core on Z270.
Intel said they did memory routing features to support DDR4-2666 MHz.

 

[Alaric]

Except low end boards regularly come without various OC options. You want high end features on a cheap chipset, yet you defend Intel forcing a high end chipset on us that likely wasn't necessary? I'm lost.

I can careless about any Z boards VRM's, they all have to run within Intel's specifications

Which historically have been at stock frequencies. And Skylake was the first semi officially supported overclocker, at least as far as Intel's specs go. There are eleven chipsets that support Socket 1151, and H110 boards don't give you the OC options Z170 boards do. Probably has something to do with power delivery/OC options, among other things. So that excuse just doesn't add up.

 

[EarthDog]

Intel said they did memory routing features to support DDR4-2666 MHz.

Repeating yourself without additional support does not make it so.

Let me ask you a question. Do you think its more likely that all z270 boards support 2666 or do not? Look at their specs... they go to 4k+. Why would you think any of them would not support 2666? Where is the logic fail?

I think you are being a bit too literal with their words.

Good news is that i will be reviewing z370 motheeboards soon. So ill be able to see these problems first hand... though, im 0-1 so far. The nzxt board was fine to 5ghz wirh 8700k.

 

[wingman99]

For that, build it on a Z170 chipset instead?

Don't recall claiming that. I think I said

And honestly, if you think the engineers have a say equal to the accountants and marketing, I would disagree. Two of my uncles were engineers, and good at it. And they would rather have invented Star Trek's transporter than "kludge" new spark plugs in a car. They wanted a clean sheet of paper for everything, even if the previous application just needed a screw. A proper design from scratch is admirable. It's also easier when the customer is paying for it.

I don't claim to be an engineer, but Intel essentially had the market to themselves for a decade or more, and could (and did) do as they pleased, because there weren't any real options (according to Intel's marketing dept., anyway). Their collective actions after Ryzen seem to indicate a rudderless ship. Again, that's how it appears from the outside. And Intel's corporate history in the CPU arena is rife with examples of the marketing department making up revenue the engineers couldn't pull off. I acknowledge I know far less about the details than Intel. That doesn't mean I'm going to blindly follow Big Blue off a personal finance cliff because "the engineers said so".

Your using a Intel product now, what are you saying? The Intel engineers allow you to overclock well, are you happy with intel? If your not use AMD for overclocking and have fun. Here is one 3335 page thread I have been following from the beginning with AMD Ryzen overclocking problems. http://www.overclock.net/forum/showthread2.php?p=26698001#post26698001

I think folks with Intel overclocking are to comfortable because the administration and engineers get it done well for everyone, almost perfect.

This is a bout the enthusiast Z boards, the cheaper main stream and value non overclocking six cores boards will come out later from what I understand.

 

[wingman99]

Except low end boards regularly come without various OC options. You want high end features on a cheap chipset, yet you defend Intel forcing a high end chipset on us that likely wasn't necessary? I'm lost.



Which historically have been at stock frequencies. And Skylake was the first semi officially supported overclocker, at least as far as Intel's specs go. There are eleven chipsets that support Socket 1151, and H110 boards don't give you the OC options Z170 boards do. Probably has something to do with power delivery/OC options, among other things. So that excuse just doesn't add up.

The new chip set is necessary for me because Intel engineers said so.

Intel's stock Z power limit specification are really generous on the Z170 Z270, not on the Z370 when overclocking running prime95. Intel's maximum watts specification for coffee lake is 209w and kaby lake 152w and My i5 pulls 124w at 4.8GHz prime95 small fft.

Repeating yourself without additional support does not make it so.

Let me ask you a question. Do you think its more likely that all z270 boards support 2666 or do not? Look at their specs... they go to 4k+. Why would you think any of them would not support 2666? Where is the logic fail?

I think you are being a bit too literal with their words.

Good news is that i will be reviewing z370 motheeboards soon. So ill be able to see these problems first hand... though, im 0-1 so far. The nzxt board was fine to 5ghz wirh 8700k.

Did you read post 22, it is so.

 

[Alaric]

Skylake is not a stellar overclocker. There was grumbling about that when it was released, along with the memory problems. Ryzen doesn't have overclocking problems, it has limits. just about anyone can get 3.9 GHz, on air, from a 3.2 GHz Ryzen. Following that pattern, 4.9 GHz Skylakes should be commonplace. As for Intel's "almost perfect administration and engineers", I remember too clearly overheating crap Pentiums getting their butts kicked by Athlons with slower clock speeds while their marketing people were strong arming Dell, HP, etc. to not use AMD chips. Then they had their coders rig the compilers for false results on AMD chips. Intel has a recent history of foisting off second place products and lying to cover it up, so I'm not just going to take their word for it.

Yes, I have an Intel chip, and it's a beast for my purposes. In fact, it trashes a stock 7700k in every respect and does it on a z170 chipset. The z270 was a collusion between M$ and Intel to get people on W10 in return for a tacit endorsement of said chip company, plain and simple.

 

[EarthDog]

so... are you saying z270 doesnt support 2666 sticks? Im not really sure what you are getting at.

 

[wingman99]

Skylake is not a stellar overclocker. There was grumbling about that when it was released, along with the memory problems. Ryzen doesn't have overclocking problems, it has limits. just about anyone can get 3.9 GHz, on air, from a 3.2 GHz Ryzen. Following that pattern, 4.9 GHz Skylakes should be commonplace. As for Intel's "almost perfect administration and engineers", I remember too clearly overheating crap Pentiums getting their butts kicked by Athlons with slower clock speeds while their marketing people were strong arming Dell, HP, etc. to not use AMD chips. Then they had their coders rig the compilers for false results on AMD chips. Intel has a recent history of foisting off second place products and lying to cover it up, so I'm not just going to take their word for it.

Yes, I have an Intel chip, and it's a beast for my purposes. In fact, it trashes a stock 7700k in every respect and does it on a z170 chipset. The z270 was a collusion between M$ and Intel to get people on W10 in return for a tacit endorsement of said chip company, plain and simple.

Ryzen does have memory overclocking problems and so many other problems I just don't have the time to tell, it's a unfinished product I would not buy. Take a look at my thread link that I have been following everyday from the beginning of Ryzen.

so... are you saying z270 doesnt support 2666 sticks? Im not really sure what you are getting at.

The motherboard manufactures support overclocking 2666 on z270. On coffee lake Intel designed memory routing features to officially support DDR4-2666 MHz on all the 6 core processors without overclocking memory. The 4 core coffee lake officially supports 2400 still.

https://ark.intel.com/products/126687/Intel-Core-i5-8400-Processor-9M-Cache-up-to-4_00-GHz

 

[trents]

Pretty sure Skylake could handle DDR4-2666 MHz. In fact, I know at least one guy who is running his DDR4 memory @3733 MHz. LOL. I'm going to oversimplify here, I'm sure, but if AMD can have a socket that came out for the Phenom and Athlon 64 chips and it supports a FX 9590, Intel should have been able to handle the bump to 6 cores on the Skylake socket.

But did it do an adequate job of supporting the FX-9590? You might remember how there was a constant stream of people wanting help with socket AM3+ using those CPUs and invariably it all boiled down to either overheating or inadequate power delivery or both on that platform. AMD was trying to squeeze out another peformance increment offering on a socket platform that just wasn't up to it.

 

[Alaric]

That was on motherboard manufacturers, though. The AM3+ socket, properly implemented for the FX chips, would run everything from an Athlon 64 to a FX 9590. Asus was the only company to build boards up to the task, but that isn't a shortcoming of the socket design. And Asus' top of the line boards are generally the industry wide top of the line, although my $160 Asus handled every FX I tried in it-but I only went to an FX 8350.

The motherboard manufactures support overclocking 2666 on z270. On coffee lake Intel designed memory routing features to officially support DDR4-2666 MHz on all the 6 core processors without overclocking memory. The 4 core coffee lake officially supports 2400 still.

G.Skill, Corsair, Gigabyte, Asus, MSI, Asrock, ad nauseum support 4000+ MHz memory on z170 and z270. Just because the JEDEC standard was 2133 MHz doesn't mean the industry sat around wringing their hands waiting for Intel to bless them with their approval of fast RAM. DDR4 2666 MHz isn't even the fastest DDR3. LOL

 

[wingman99]

That was on motherboard manufacturers, though. The AM3+ socket, properly implemented for the FX chips, would run everything from an Athlon 64 to a FX 9590. Asus was the only company to build boards up to the task, but that isn't a shortcoming of the socket design. And Asus' top of the line boards are generally the industry wide top of the line, although my $160 Asus handled every FX I tried in it-but I only went to an FX 8350.



G.Skill, Corsair, Gigabyte, Asus, MSI, Asrock, ad nauseum support 4000+ MHz memory on z170 and z270. Just because the JEDEC standard was 2133 MHz doesn't mean the industry sat around wringing their hands waiting for Intel to bless them with their approval of fast RAM. DDR4 2666 MHz isn't even the fastest DDR3. LOL

Call for support when overclocking memory at the rated speed and see what guarantee you get. Do you feel lucky.

Sky lake is only 14nm, kaby lake is 14 nm+. Sky lake clocks exactly like it should for no die process change like kaby lake had. My Z170 with sky lake clocked perfect 3200 14-14-14-34 memory speed with only using XMP.