Skylake owners thread

[Khato]

Max operating voltage is stated at 1.52v, way above stock.

Note that maximum operating voltage specified by Intel in the datasheet has been at 1.52v since Sandybridge I believe. Seeing as how Lynnfield shows 1.55v, and Penryn core 2 duo is 1.45v... Also, just because the datasheet says that such is the 'maximum' doesn't mean that it's safe or recommended - last I checked you aren't going to be doing anything at the minimum recommended 0.25v.

Did you even read the Anandtech review...?

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/6

I wouldn't necessarily put much stock in such given that 'stock' voltage is being interpreted by the motherboard. My impression is that early release BIOS tended to be a bit excessive in this regard? (aka, look at the huge variances in load power consumption between some different boards at 'stock' when idle power is comparable.) Would be interesting to hear current reports in this regard.

 

[Deders]

Well for what it is worth, my z170 Deluxe was around 1.3v at stock, depending on load, and it has been confirmed by Asus that the reading in Aida matches up with what they measure from the socket with a multimeter.

The Aida reading is the same as the HWInfo reading and the AI suite reading for Vcore.

 

[crashtech]

1.3V is enough to overwhelm a stock HSF even down around 4.0GHz, maybe that's partly why they don't include them anymore.

 

[LTC8K6]

1.3V is enough to overwhelm a stock HSF even down around 4.0GHz, maybe that's partly why they don't include them anymore.

I think they intended to sell us these:

https://www.ple.com.au//imagelibrary/inventoryitemimages/68978-620292-full.jpg

https://www.ple.com.au//imagelibrary/inventoryitemimages/67622-620292-full.jpg

Which we probably wouldn't want, but they are nowhere to be found, anyway.

 

[DrMrLordX]

Huh. All these motherboards overvolting the hell out of Skylake by default makes me think of Kaveri refresh . . .

 

[crashtech]

I am unable to keep the 6600K under 100C testing at 4.7GHz and above with voltages over 1.4V. I am wondering if there is either something wrong with the Corsair H110, or the TIM under the lid is crappy.

According to this chart, the H110 should be able to dissipate 340W without excessive temp rise:

Though honestly, all those numbers seem (overly?) optimistic to me.

 

[Edrick]

I am unable to keep the 6600K under 100C testing at 4.7GHz and above with voltages over 1.4V. I am wondering if there is either something wrong with the Corsair H110, or the TIM under the lid is crappy.

According to this chart, the H110 should be able to dissipate 340W without excessive temp rise:


Though honestly, all those numbers seem (overly?) optimistic to me.

I have a Corsair H100i GTX and at 1.3v and 4.4ghz, my 6600k jumps into the high 80's running Prime95. Yet at 4.3ghz and 1.26v, my temps stay in the mid 70's running the same tests. I even replaced the corsair TIM with Arctic MX4. So that appears to be the sweet spot for my chip. However, I have no doubt that a delid would greatly improve temps.

 

[Deders]

I am unable to keep the 6600K under 100C testing at 4.7GHz and above with voltages over 1.4V. I am wondering if there is either something wrong with the Corsair H110, or the TIM under the lid is crappy.

According to this chart, the H110 should be able to dissipate 340W without excessive temp rise:

Though honestly, all those numbers seem (overly?) optimistic to me.

What are you testing with? what mobo do you have,? what is your VID in HWinfo? Are you using offset, manual or adaptive voltage? what about LLC?

I'm not sure where they get their 340W load figure from, at 1.4v and 4.6GHz my i7 reads 96w.

At those settings my CPU will nudge 90c in small FTT Prime, but be below 70c for everything else. That is with the NH-D15 air cooler.

Using Gelid Extreme Tim, I'd expect MX4 to be close to this.

 

[Haserath]

I am unable to keep the 6600K under 100C testing at 4.7GHz and above with voltages over 1.4V. I am wondering if there is either something wrong with the Corsair H110, or the TIM under the lid is crappy.

According to this chart, the H110 should be able to dissipate 340W without excessive temp rise:

Though honestly, all those numbers seem (overly?) optimistic to me.

That's if you're getting proper heat transfer. If you put the cooler directly on a 600mm^2 GPU, it'd easily dissipate 340W with a low delta like those.

Look at the testing methodology:
http://www.anandtech.com/show/7738/closed-loop-aio-liquid-coolers/8

They created what is essentially a bare die that looks decently big(and evenly distributes heat), compared to these small skylake cores that are 30mm2? under terrible TIM under an IHS.

 

[DrMrLordX]

Haven't people been reporting drops of ~20C with Skylake delid/relid?

 

[crashtech]

Haven't people been reporting drops of ~20C with Skylake delid/relid?

Would this indicate that the current Skylake "K" models are not using the "improved" TIM ala Devil's Canyon?

@Deders, I'll have more to add once I get home this evening.

 

[Excessi0n]

Did you even read the Anandtech review...?

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/6

That seems bizarrely high. My Z170 Deluxe put my 6700K at 1.23 idle/1.27 load.

So, I'm still wrong. How many users will feel OK with their CPU over 1.4V all the time, though?

There was a guy on OCN who talked to an Intel engineer at one of the conferences, and supposedly the VCore is actually lower than is reported because the voltage is stepped down somewhere between the measuring point and the actual core. I don't know if it's true, but it would nicely explain why Skylake can tolerate such a seemingly high "VCore."

I'm not going to run mine above 1.35v though. Somebody else can be the guinea pig.

 

[Deders]

Haven't people been reporting drops of ~20C with Skylake delid/relid?

Would this indicate that the current Skylake "K" models are not using the "improved" TIM ala Devil's Canyon?

Some people report 7-10c difference, some maybe a bit more. From what I gather it has more to do with the distance between the die and IHS.

That seems bizarrely high. My Z170 Deluxe put my 6700K at 1.23 idle/1.27 load.

Pretty close to 1.3

There was a guy on OCN who talked to an Intel engineer at one of the conferences, and supposedly the VCore is actually lower than is reported because the voltage is stepped down somewhere between the measuring point and the actual core. I don't know if it's true, but it would nicely explain why Skylake can tolerate such a seemingly high "VCore."

I'm not going to run mine above 1.35v though. Somebody else can be the guinea pig.

I will be your guinea pig, I have the same mobo and the Intel tuning plan so I can get an instant replacement.

I'm happy to run mine at 1.4 but not much more. There are 2 voltage readings I can see and only HWInfo exposes them both:

Vcore has a single reading that is present in HWinfo, Aida as well as AI Suite.

Vid on the other hand has 4 readings, one for each core.

With no LLC the VID is much lower than the Vcore. When I apply some levels of LLC (I'm thinking 5 but I could be wrong) the Vcore is lower than the VID, at least with my MB/CPU combo.

I have it set to LLC level 6 and the VID is reading 0.004v lower than the Vcore and so far this has produced the most stable results with the least heat.

I've not tried LLC below 5 or above 7.

I'm stable with 8 hours of realbench but I've been having trouble with Prime. It runs a lot better now that I have v28.7 that eliminates a bug that gave errors in a situation where there shouldn't have been any.

I ran a blend last night and core 3 came up with an error less than an hour in, the rest of them stayed strong for 6 hours. The last reported test on core 3 was 15k which I presume was a small FTT.

Since then I've run just 15k for an hour without issue, and the small FTT setting for 3 hours without issue.

I then ran large FTT's and core 3 came up with an error, can't remember if it was 12xk or 144k but I'm thinking maybe it is a VTT or Vram voltage issue. Will test again soon.

 

[crashtech]

What are you testing with?

LinX 0.6.5

what mobo do you have,?

ASRock Z170 Extreme4

what is your VID in HWinfo?

1.414V

Are you using offset, manual or adaptive voltage? what about LLC?

Manual, LLC set to max which is Level 1 on this mobo.

Turns out that the AVX2 load was pushing 0.08V extra into the CPU without me paying proper attention. Apparently this happens even with a manual voltage setting, which was not the case with my last two mobos. So the voltage that was overheating the CPU was more like 1.48-1.52V, though I don't have any screencaps of it to be sure. I think the problem is under control now, but I am still getting 99°C @ 1.44V (1.36 + 0.08 with AVX2) at 4.7GHz. I have remounted the CPU twice, contact is good. Also, I'm not quite sure what I did differently today, but VID is maxing at 1.115 right now

 

[Deders]

I am unable to keep the 6600K under 100C testing at 4.7GHz and above with voltages over 1.4V. I am wondering if there is either something wrong with the Corsair H110, or the TIM under the lid is crappy.

LinX 0.6.5 ASRock Z170 Extreme4 1.414V Manual, LLC set to max which is Level 1 on this mobo.

Turns out that the AVX2 load was pushing 0.08V extra into the CPU without me paying proper attention. Apparently this happens even with a manual voltage setting, which was not the case with my last two mobos. So the voltage that was overheating the CPU was more like 1.48-1.52V, though I don't have any screencaps of it to be sure. I think the problem is under control now, but I am still getting 99°C @ 1.44V (1.36 + 0.08 with AVX2) at 4.7GHz. I have remounted the CPU twice, contact is good. Also, I'm not quite sure what I did differently today, but VID is maxing at 1.115 right now

Just tried a Linx run and my temps do reach 95c on one core, the rest not far behind.

I had to set my LLC so that my Vcore was about the same or just a bit higher than my VID at the given frequency (4.6) otherwise my Vcore would be giving off too much heat but the cores (vid) wouldn't be getting nearly enough voltage to be stable.

It's hard to say with other boards as I have no idea how well they are calibrated. 1.115 is suspiciously low, it can't be accurate.

I've redone all 3 prime tests with a raised VCCIO and they passed for over 8 hours. Vid reads 1.404, Vcore reaches 1.424 under these extreme loads, but 1.408 for normal use.

Had to raise the VCCIO from 1.12 (auto) to 1.1375 (set in bios) and now the bios reads 1.16.

 

[Spydermag68]

I am almost done with my 6700k build. I need to get a few 140mm fans for the front case and the top. The current front fan only has 3 pins. I am also waiting on a longer HD audio cable from Corair. The case is a 300R and they have known about the cable since 2012. I am glad that they are sending it to me for free.

Build:
CPU: i7-6700k
PS: EVGA 750 P2
MB: Gigabyte z170-UD5
Memory: Corsair 3000, 16 gigabytes
Case: Corsair 300R

For the old computer see my sig.

 

[Deders]

Is it up and running? do you intend to overclock?

 

[crashtech]

Benchmarks!

 

[Deders]

Benchmarks!

or it didn't happen!!

 

[Spydermag68]

It is not up and running I have had very limited time to put it together at night. One night I put the CPU, memory and CPU cooler on and installed it into the case. The next night I noticed that I forgot to take off the plastic shield off the CPU cooler so it are up all of my time pulling it all out and redoing it. The next night I was routing cables and doing a bad job of it and that is when I found out about the short HD audio cable and the cheap 3-pin front panel fan. Off to microcenter tomorrow for those.

For a while I will not over clock until I find it stable at stock.

 

[LTC8K6]

A 3 pin voltage controlled fan is fine for a case fan.

You will need to tell your mobo that it's a voltage controlled fan rather than a PWM fan.

 

[Idontcare]

It is not up and running I have had very limited time to put it together at night. One night I put the CPU, memory and CPU cooler on and installed it into the case. The next night I noticed that I forgot to take off the plastic shield off the CPU cooler so it are up all of my time pulling it all out and redoing it. The next night I was routing cables and doing a bad job of it and that is when I found out about the short HD audio cable and the cheap 3-pin front panel fan. Off to microcenter tomorrow for those.

For a while I will not over clock until I find it stable at stock.

What!?

This might suggest that you consider having someone assist you in your build

 

[Spydermag68]

I only have 4-pin connection on the motherboard. The 3-pin will not fit. I am not connecting anything that needs the old molex power so the 3-pin splitter to 5 volts is not going into the case.

 

[Spydermag68]

What!?

This might suggest that you consider having someone assist you in your build

Give me a break. It was a mistake and at least I corrected it before I fired up the computer. I have been building computers since I had to plan out IRQs an flipping bit switches.

 

[crashtech]

3-pin connectors WILL fit on 4-pin headers, they just won't be speed controlled.