Intel Skylake / Kaby Lake

[RichUK]

I'm bored of Skylake X, I want me some Coffee Lake 6C 14nm++ goodness.

 

[TheF34RChannel]

I'm bored of Skylake X, I want me some Coffee Lake 6C 14nm++ goodness.

I have an identical craving... But man if it's Oct - Nov for an actual launch... I am so hell bend on August. Keep in mind that I likely won't buy upon launch but rather a few weeks later to unearth any issues and get a stable BIOS.

 

[jj109]

http://www.overclock.net/t/1632870/skylake-x-kaby-lake-x-combined-discussion/200#post_26197736

Just one set of limited results, but the mesh/L3/RAM memory latency impact is huge for Skylake X.

 

[coercitiv]

I may eat my words, but the upcoming Threadripper will be great at workstation loads (will beat many Intel processors) and probably not very good at gaming (not bad, mind you, but not a 7740k killer either).

TR won't be winning any game benchmarks, that's for sure.

One problem we have around here is we lack context when discussing the "moar cores" potential in gaming: some seem to think core usage will explode like a piñata, while others (myself included) think we're just moving from 4c to 6c / 8c in an orderly, paced, predictible fashion. The same is true with skeptics, some think high speed 4c will stay on top for years to come, other are just trying to warn the piñata crowd.

 

[TheF34RChannel]

Read this please regarding the VRM: https://forum.overclock3d.net/showpost.php?p=951591&postcount=18

So, as Tom knows, I tested this same PSU with a 300W load on a single EPS12V connector.

Temperatures of the connector pins never got near 40°C. Voltage only dropped .22V despite having only three 20 AWG wires going to a four pin EPS12V. With that said, I go back on what I said and will say that I would GLADLY use this PSU's, or any PSU's, EPS12V connector to push 300W of power.

I don't doubt the guy believes his results... but something is wrong with either a piece of his kit or his thermometer.

 

[JoeRambo]

http://www.overclock.net/t/1632870/skylake-x-kaby-lake-x-combined-discussion/200#post_26197736

Just one set of limited results, but the mesh/L3/RAM memory latency impact is huge for Skylake X.

It seems like Skylake-X will love Uncore/Memory OC. Those 25% in TW: Warhammer are amazing.

 

[Arachnotronic]

http://www.overclock.net/t/1632870/skylake-x-kaby-lake-x-combined-discussion/200#post_26197736

Just one set of limited results, but the mesh/L3/RAM memory latency impact is huge for Skylake X.

It's clear that the beast needs to be fed better. Hopefully, with the 14nm++ refresh (Cascade Lake) Intel ships these things with faster L3 cache and higher out-of-the-box memory speed support.

 

[Arachnotronic]

Read this please regarding the VRM: https://forum.overclock3d.net/showpost.php?p=951591&postcount=18

Wow, so basically der8auer spread a lot of FUD for no reason.

And that, my dear friends, is why a sample size of one isn't enough to make any reliable conclusions.

 

[coercitiv]

And that, my dear friends, is why a sample size of one isn't enough to make any reliable conclusions.

Don't know about that, it sure seems like a sample of one was enough to put der8auer in the FUD section.

The sad truth is those VRM heatsinks have just an esthetic purpose, which is quite pathetic considering the Enthusiast moniker of the platform. Shrouds and LEDs boys and girls!

This was probably just a business move on the part of der8auer, to promote his upcoming VRM delid tool. /s

 

[frozentundra123456]

Wow, so basically der8auer spread a lot of FUD for no reason.

And that, my dear friends, is why a sample size of one isn't enough to make any reliable conclusions.

Yea, but it gets him a lot of clicks. It is the old fart in me I guess, but I always take youtube videos with a large grain of salt.

 

[TheF34RChannel]

It's clear that the beast needs to be fed better. Hopefully, with the 14nm++ refresh (Cascade Lake) Intel ships these things with faster L3 cache and higher out-of-the-box memory speed support.

The low uncore frequency compared to the core frequency has me surprised in all honesty! You'd think they'd know it needed to be much higher to be beneficial (they probably did know). Do you think Intel will stay with current cache size models or go back to < L2 & > L3? Can they even go back with the mesh architecture?

Wow, so basically der8auer spread a lot of FUD for no reason.

And that, my dear friends, is why a sample size of one isn't enough to make any reliable conclusions.

Yeah seems that way. JonnyGuru has been making more posts in several OC3D forum threads in which he elaborates a bit; too many to look for and link, however I'm sure he has them on his own site/channel/whatever as well.

 

[jpiniero]

Do you think Intel will stay with current cache size models or go back to < L2 & > L3? Can they even go back with the mesh architecture?

I'm sure the L2 and L3 will be tweaked. The mesh is staying though, it's got to be a prerequisite for EMIB.

 

[TheF34RChannel]

I'm sure the L2 and L3 will be tweaked. The mesh is staying though, it's got to be a prerequisite for EMIB.

How could I not think of that (regarding the mesh)! I know the EMIB basics but would like to know more; do you have any sites you can recommend me?

 

[Jan Olšan]

Wow, so basically der8auer spread a lot of FUD for no reason.

You guys should cut him some slack. A week ago, when he promised those great overclocks with i9-7900X, remember that people were reposting his stuff gladly.
And on the particular boards he criticised, he didn't fake VRM temperature measurements or something like that, did he?

 

[KompuKare]

Yea, but it gets him a lot of clicks. It is the old fart in me I guess, but I always take youtube videos with a large grain of salt.

Er, I'm sure the reason Caseking.de (and overclockers.co.uk which is also owned by Caseking but has 8pack instead of der8auer) employ these top overclockers is so that they can sell more stuff and especially their picked overclock bundles. And Skylake-X is a high priced item which is exactly what they like to sell there.
der8bauer coming out and saying people should hold off on LGA2066 boards for now is not going to do wonders for their sales and a few internet clicks are not going compensate for that.

 

[TheF34RChannel]

This went live just now!

 

[Ajay]

Wow, so basically der8auer spread a lot of FUD for no reason.

And that, my dear friends, is why a sample size of one isn't enough to make any reliable conclusions.

I don't know about FUD, but he just lost allot of credibility, IMO.

 

[twothreefive]

I don't know about FUD, but he just lost allot of credibility, IMO.

I agree. I understand there is a lot of pent up Intel hatred but I have to wonder how much AMD is paying him.

 

[ZGR]

I agree. I understand there is a lot of pent up Intel hatred but I have to wonder how much AMD is paying him.

I wouldn't go that far haha. I think he was being genuine in his video but he had other factors that caused the unusually high VRM temperatures.

I don't know anyone who hates Intel, but personally, I was baffled at the motherboard manufacturers about the VRM's when I first saw his video. It is such a relief to know that it is a non issue.

It probably was his PSU.

 

[Jan Olšan]

Wait, the PSU can cause high temps on its connectors and cables if they are too thin or too little wires/bad connectors. But how could the PSU cause higher VRM (voltage regulators on the board) temps?

BTW the TR4 boards seem to have same style of VRM "decoration" heatsinks, so not sure if the boards made for AMD will be any better.

 

[jj109]

Wait, the PSU can cause high temps on its connectors and cables if they are too thin or too little wires/bad connectors. But how could the PSU cause higher VRM (voltage regulators on the board) temps?

If the connectors and wires are hot, that means there's power being dissipated over the cable and voltage is dropping across it. If the voltage drop is significant, the VRMs will need to pull more current to sustain its output voltage.

So the question now is why is der8auer's cable so hot when other people don't see the issue.

 

[TheF34RChannel]

If the connectors and wires are hot, that means there's power being dissipated over the cable and voltage is dropping across it. If the voltage drop is significant, the VRMs will need to pull more current to sustain its output voltage.

So the question now is why is der8auer's cable so hot when other people don't see the issue.

There's obviously something wrong with either his PSU or cable

 

[Jan Olšan]

Again, that can't increase temperature of his onboard VRM significantly. The cable was a separate problem, caused likely by the fancy PSU having some silly special connectors to look pretty.

The VRMs overheating is a different issue, and could have only been caused by inadequate cooling or too high power losses, IMHO. At 300W CPU consumption, it is not weird for those small heatsinks that don't have proper fins to be inadequate.

 

[TahoeDust]

Again, that can't increase temperature of his onboard VRM significantly. The cable was a separate problem, caused likely by the fancy PSU having some silly special connectors to look pretty.

The VRMs overheating is a different issue, and could have only been caused by inadequate cooling or too high power losses, IMHO. At 300W CPU consumption, it is not weird for those small heatsinks that don't have proper fins to be inadequate.

Why do you think no one been able to replicate his results?

 

[moonbogg]

Why do you think no one been able to replicate his results?

Silicon lottery had the same issues with VRM temps. It wasn't a sample size of one. It was two highly credible sources telling everyone the same thing about VRM temps. Maybe there are VRM cooling issues with some boards. Maybe others are fine. There could have been some manufacturing inconsistency due to the rushed nature of the release. We won't actually know the truth for another month or so when tons of people report their own findings. The Intel hate bandwagon is exciting and all, but I hope this temp issue was just a misunderstanding or mistake. We need competition and solid products from both camps. That's the only way we all win.