Intel Skylake / Kaby Lake

[Arachnotronic]

Tahoe, what RAM kit are you using?

 

[pantsaregood]

The 8-pin connector(s) are only one minor part of the discussion, the more relevant issue is how many pins on the socket provide power, and whether it's feasible to share their maximum current carrying capacity with two more cores without creating other problems.

You're right, but this ignores the fact that running Skylake/Kaby Lake at >5.0 GHz can easily push power consumption to >50% of stock consumption without issues with the socket delivering power. High-end motherboards can push Skylake/Kaby Lake to >7.0 GHz, which - without factoring in voltage increases - represents at least a 66% increase in power consumption.

Given that high end "overclocking" motherboards use the same LGA 1151 socket and that socket is capable of running at >7.0 GHz, we can deduce that the socket is fully capable of delivering sufficient power.

 

[TahoeDust]

Tahoe, what RAM kit are you using?

g.skill F4-3600C16Q-32GTZR
https://www.newegg.com/Product/Product.aspx?Item=N82E16820232492

 

[crashtech]

You're right, but this ignores the fact that running Skylake/Kaby Lake at >5.0 GHz can easily push power consumption to >50% of stock consumption without issues with the socket delivering power. High-end motherboards can push Skylake/Kaby Lake to >7.0 GHz, which - without factoring in voltage increases - represents at least a 66% increase in power consumption.

Given that high end "overclocking" motherboards use the same LGA 1151 socket and that socket is capable of running at >7.0 GHz, we can deduce that the socket is fully capable of delivering sufficient power.

I don't disagree, but clearly the scenarios you mention don't have anything to do with complying with specs. Pushing a hexacore to the limit on air with the current socket's power delivery arrangement might just be a sketchy proposition. Most of us just don't know.

 

[Ajay]

I don't disagree, but clearly the scenarios you mention don't have anything to do with complying with specs. Pushing a hexacore to the limit on air with the current socket's power delivery arrangement might just be a sketchy proposition. Most of us just don't know.

We'll all have to wait for reviews which, according to rumors, is going to be awhile

 

[Arachnotronic]

I don't disagree, but clearly the scenarios you mention don't have anything to do with complying with specs. Pushing a hexacore to the limit on air with the current socket's power delivery arrangement might just be a sketchy proposition. Most of us just don't know.

I wouldn't worry about the socket, but I would be worried about some of the cheaper Z170 boards not having robust enough power delivery systems.

 

[crashtech]

I wouldn't worry about the socket, but I would be worried about some of the cheaper Z170 boards not having robust enough power delivery systems.

I've been trying to give Intel the benefit of the doubt, but it sounds like you think "1151v2" is most likely a money grab.

 

[Arachnotronic]

I've been trying to give Intel the benefit of the doubt, but it sounds like you think "1151v2" is most likely a money grab.

Not a money grab for Intel (PCH revenue from enthusiast desktop is really peanuts & losing out on CPU sales because people don't want to buy new mobos or being pissed off to say "I'm going Ryzen because AMD supports its sockets" could be a net loss for Intel), but a kiss blown to the motherboard makers, probably to help make up for the fact that they have to do a whole new round of boards for a pulled-in Coffee Lake to deal with Ryzen.

 

[jpiniero]

I still feel like they changed enough that it's more than just more power.

Not a money grab for Intel (PCH revenue from enthusiast desktop is really peanuts & losing out on CPU sales because people don't want to buy new mobos or being pissed off to say "I'm going Ryzen because AMD supports its sockets" could be a net loss for Intel), but a kiss blown to the motherboard makers, probably to help make up for the fact that they have to do a whole new round of boards for a pulled-in Coffee Lake to deal with Ryzen.

If anything the mobo makers are the big losers since they will have loads of soon to be unsellable Z270 boards.

 

[Arachnotronic]

I still feel like they changed enough that it's more than just more power.

I don't think so, there is plenty of evidence of CFL working in Z270 boards.

If anything the mobo makers are the big losers since they will have loads of soon to be unsellable Z270 boards.

Mobo makers have known about this pull-in for quite some time, so they've had plenty of opportunity to manage their inventories appropriately. A lot of these mobo makers' board designs are pretty much based off of reference designs that Intel develops, so it's not as though they didn't know that this was coming.

 

[Wyrm]

I would really love to get 100GB/s memory and your Read/Write stats seem to achieve that. However, when I compare your memory benchmark to Arachnotronic's, I can see that Copy did not scale as well as Read/Write. He had 82GB/s with DDR4-3000 and you have 86GB/s with DDR4-3600. Do you happen to know why it improved so little with the higher memory clock?

 

[frozentundra123456]

Not upset at all. Just pointing out the ridiculous claim you made with zero facts to back it up. But seeing how you are the only person on these forums who is never wrong, who am I to argue.

I will make sure to check Intel's HEDT sales after the 8700K gets released. I should see a 90% drop in sales after that according to you.

Weird, wonder why the single thread is slower than the 7700K.

They are very close.

 

[Edrick]

240mm AIO.

I am not even close to getting these numbers. What is your Cinebench score?

 

[Arachnotronic]

I am not even close to getting these numbers. What is your Cinebench score?

As in, you're much lower or much higher?

 

[Edrick]

As in, you're much lower or much higher?

Much lower. And I have faster RAM than you.

 

[Arachnotronic]

Much lower. And I have faster RAM than you.

Update your BIOS

 

[pantsaregood]

I still feel like they changed enough that it's more than just more power.



If anything the mobo makers are the big losers since they will have loads of soon to be unsellable Z270 boards.

It doesn't look like anything has really been changed with the PCH - Z370 looks to be just a Z270 rebrand. Z390 is going to be the new PCH. Socket power delivery looks like it is the only plausible explanation (and that's stretching it since people are pushing Kaby Lake to >7.0 GHz on LGA 1151 already), so no - there's no greater difference.

Motherboard manufacturers are unlikely to take a hit. I'm expecting Z370 boards to be Z270 boards with Z370 attached.

 

[CakeMonster]

Z390 is a new PCH, with some features like native USB 3.1 and some integrated connectivity stuff manufactured in 14nm rather than 22nm, so should be slightly more efficient, too.

But, it's really not a big deal.

Wish we had a date though. If I don't have to wait too long, it would be nice to get Z390.

 

[nvgpu]

Cannon Lake PCH Z390 chipset will probably launch at CES 2018 January 9-12 together with the rest of the Coffee Lake SKUs that didn't launch in October. When motherboards will be available is anyone's guess.

 

[mikk]

I don't think so, there is plenty of evidence of CFL working in Z270 boards.

What evidence? I haven't seen a single evidence. Or do you mean the KBL misread entries with CFL due to the KBL PCH? This is no evidence because we know that Z370 uses KBL PCH.

 

[nvgpu]

https://downloadcenter.intel.com/download/27058/Graphics-Intel-Graphics-Driver-for-Windows-15-46-

iKBLULTGT2R = "Intel(R) UHD Graphics 620"

Intel 22.20.16.4771 driver for Kaby Lake-R, Kaby Lake & Skylake.

 

[phillyman36]

The 8700k combined with the Z370 has 24 lanes right? (just like z270 and 7700K pci lane talk is where i get confused) I have 1 graphics card, 2 nvme ssds, and 1 sata ssd. The nvme drives should run at full speed right?

 

[jpiniero]

The 8700k combined with the Z370 has 24 lanes right? (just like z270 and 7700K pci lane talk is where i get confused) I have 1 graphics card, 2 nvme ssds, and 1 sata ssd. The nvme drives should run at full speed right?

No. IIRC, It's the same as Kaby - 16 lanes on the processor and 24 on the chipset. But to go from the chipset to the processor is only x4.

 

[eddman]

No. IIRC, It's the same as Kaby - 16 lanes on the processor and 24 on the chipset. But to go from the chipset to the processor is only x4.

I don't know enough about this subject. I suppose a fast chipset-CPU connection is important when loading applications and data from storage into RAM. In what other situations it plays a major role?

Copying data between storage devices should be direct, correct?

 

[Bouowmx]

RAM hasn't been connected to Intel chipsets since 2008.

DMI 3.0, or PCI Express 3.0 x4, has bandwidth ~4 GB/s per direction. Fine for most users, but not enough for multiple NVME drives.

Spoiler: Intel Z270 block diagram