[Techspot] BCLK overclocking non-K Intel Skylake CPUs is now possible, tested here

[VirtualLarry]

With the cheapy celerons you like to buy, I'm surprised you would be worried.

LOL. I'm not made of money. Just because I buy mostly low-end stuff, doesn't mean that I'm going to let them go out "in a blaze of glory".

 

[crashtech]

Hmm, after watching a friend's 45nm C2D Pentium dual-core degrade at 1.425v BIOS, I'm not really sure I want to "juice" a 14nm chip that much.

Edit: Didn't Intel lower either TJmax or TcaseMax for Skylake? As in, it's a good idea to keep them running cooler than Haswell?

I didn't see heat start becoming a problem until over 1.4V, and that was with a 6600K. Lack of FIVR also means some heat producing parts are not on-die anymore. Nearly everyone is using more than 1.3V:

http://www.anandtech.com/show/9533/intel-i7-6700k-overclocking-4-8-ghz

http://www.tweaktown.com/guides/748...-intel-skylake-overclocking-guide/index5.html

http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics

http://www.tomshardware.com/reviews/skylake-intel-core-i7-6700k-core-i5-6600k,4252-3.html

My guess is that you will see good improvements with some modest voltage increases, from my experience, Skylake likes voltage despite it's diminutive size. Some Skylake motherboards run as much as 1.39 to a stock CPU under heavy load.

 

[VirtualLarry]

My guess is that you will see good improvements with some modest voltage increases, from my experience, Skylake likes voltage despite it's diminutive size. Some Skylake motherboards run as much as 1.39 to a stock CPU under heavy load.

1.39 on a stock Skylake? That sounds... insane... to me. No?

 

[crashtech]

1.39 on a stock Skylake? That sounds... insane... to me. No?

It is what it is. I've read some speculation that the cores might not actually be seeing those values, but I don't think many people know for sure, the ones that do aren't talking. It's obvious from reading about and working with a couple Skylakes that they need more voltage than Haswell to run, probably around 0.1V more on average to achieve similar clocks.

Maybe you could turn it up temporarily, you may find several hundred more MHz for just a few hundredths of a volt more, never know til you try.

 

[Excessi0n]

Hmm, after watching a friend's 45nm C2D Pentium dual-core degrade at 1.425v BIOS, I'm not really sure I want to "juice" a 14nm chip that much.

Well, for whatever reason, Skylake and its 14nm process seems to love voltage. The Vcore maximum listed in the Skylake-family datasheet is a staggering 1.52 volts. D:

Personally, I'm quite happy running mine at a "mere" 1.35...

 

[Bubbleawsome]

Well, for whatever reason, Skylake and its 14nm process seems to love voltage. The Vcore maximum listed in the Skylake-family datasheet is a staggering 1.52 volts. D:

Personally, I'm quite happy running mine at a "mere" 1.35...

Can you source that? If it's true I'd love a skylake chip. I can't find anything like that for any chip families though.

 

[ZGR]

How much can Skylake be undervolted at stock and OC speeds? Just curious.

 

[VirtualLarry]

Ok, now I've got my first G4400 CPU at 4.4Ghz and 1.350V BIOS. CPU-Z reads 1.312V.

At 1.350V BIOS, I tried 4.455Ghz (135.0 BCLK), and I was able to get a CineBench 11.5 screenshot, but it wasn't stable under OCCT load, STOP 0x101 BSOD. 4.401Ghz seemed OK.

Edit: Wholly Moley! My UPS beeped and my PC shut off. I installed my UPS software, and at IDLE, it's drawing 221W! Subtract 60W for the LCD monitor, and that's what the PC is drawing.

My other non-overclocked G4400 sips 59W under full load, WITH the monitor. (Different model monitor than this rig.)

With the CPU crunching, and the 7950, it's drawing nearly 350W! A dual-core!

I hope I didn't degrade my CPU, and that's why it's suddenly taking a lot of juice.

 

[crashtech]

Can you source that? If it's true I'd love a skylake chip. I can't find anything like that for any chip families though.

http://www.intel.com/content/dam/ww...sktop-6th-gen-core-family-datasheet-vol-1.pdf

Page 114.

 

[crashtech]

Ok, now I've got my first G4400 CPU at 4.4Ghz and 1.350V BIOS. CPU-Z reads 1.312V.

At 1.350V BIOS, I tried 4.455Ghz (135.0 BCLK), and I was able to get a CineBench 11.5 screenshot, but it wasn't stable under OCCT load, STOP 0x101 BSOD. 4.401Ghz seemed OK.

Not too shabby!

 

[Burpo]

I hope I didn't degrade my CPU, and that's why it's suddenly taking a lot of juice.

"Each processor is programmed with a maximum valid Voltage indentification value (VID) that is set at manufacturing and cannot be altered. Individual maxmimum VID values are calibrated during manufacturing, such that 2 processors at the same frequency may have different settings within the VID range"

Operating Voltage: Minimum 0.55v, Maximum 1.5v

page 90
5.1.5.8 Thermtrip# Signal
Regardless of Enabling the automatic or on-demand modes, in the event of a catastrophic cooling failure, the package WILL automatically shut down when the silicon has reached an elevated temperature that risks physical damage to the product. At this point the THERMTRIP# signal WILL go active.

Relax Larry, You haven't "degraded" your processor You haven't even really pushed it yet..

 

[VirtualLarry]

Relax Larry, You haven't "degraded" your processor You haven't even really pushed it yet..

Oh really? Then why is my (now non-OCed) G4400 rig taking nearly 200W at idle, and 450W+ at full CPU + GPU load?

My UPS maxes out at 330W.

And I was successfully crunching at full load under that wattage for a day or so.

So I think I blew a VRM, or something in my PSU, because the power that I'm drawing at the wall now (after the UPS overload and shutdown incident) is insane.

 

[Burpo]

What is the draw without the 7950?

 

[crashtech]

A dual core Skylake won't be the cause of current draw like that even if it is on fire.

 

[VirtualLarry]

Ok, I changed out the (1-ish year old) EVGA 500B 500W PSU, for an EVGA 600B 600W PSU. (It's what I had immediately handy.)

Idle wattage is down to 128W, load with CPU crunching is 148W, load with CPU + GPU crunching is 312W.
(This is at stock CPU / GPU clocks.)

Much better than 211W and 450W+!

The software for my UPS that I installed, was the newest version 1.61. But I noticed that after the incident, it was reporting that my UPS max wattage was 210W. (It's supposed to be 330W.)

I copied over the original 1.50 version of the software that I used on my other rig, and uninstalled 1.61, rebooted, and installed 1.50, and the UPS control panel still shows a max wattage of 210W. Very weird. There's no user configuration for that setting, either.

So, something fried part of my UPS when it overloaded too, along with my PSU.

Still, not sure that was the problem alone, if my idle wattage is 128W. Though, I measured the monitor's wattage at around 89-99W. So possibly that's right.

 

[SPBHM]

it would be interesting to see the default idle power usage (with power saving stuff turned on) vs the bclk OC.

 

[VirtualLarry]

it would be interesting to see the default idle power usage (with power saving stuff turned on) vs the bclk OC.

Well, remember, SpeedStep and the power-saving features get turned off when you do a BCLK OC.

 

[SPBHM]

Well, remember, SpeedStep and the power-saving features get turned off when you do a BCLK OC.

that's why I'm curious, how much more Skylake will consume at low load with full voltage/clock

I remember it not making a huge difference with my Core 2 CPUs, but since then I never really used a PC without those enabled anymore.

 

[Shmee]

Is this only a thing on ASRock boards? If so, good for them then. Also, does it support the 6700T?

 

[Burpo]

Oh really? Then why is my (now non-OCed) G4400 rig taking nearly 200W at idle?

Well, remember, SpeedStep and the power-saving features get turned off when you do a BCLK OC.

Which is it? OC or Not? I'm confused..

 

[DrMrLordX]

So, something fried part of my UPS when it overloaded too, along with my PSU.

You got a spare surge protector around, or another UPS, so you can test without the defective UPS?

 

[coercitiv]

So, something fried part of my UPS when it overloaded too, along with my PSU.

And yet you choose to rely on power readings from the UPS.

The mere fact you believe that changing the PSU resulted in a drop of 90W in system power consumption is mystifying: no component in your configuration could handle an extra 90W worth of heat.

 

[VirtualLarry]

And yet you choose to rely on power readings from the UPS.

The mere fact you believe that changing the PSU resulted in a drop of 90W in system power consumption is mystifying: no component in your configuration could handle an extra 90W worth of heat.

Maybe... before it shut down, the fans were screaming away, though.

I'm wondering if the APFC circuit in the PSU blew, or something. Because it clearly wasn't running at 80Plus Bronze efficiency, if it was truly drawing 464W at the wall at the UPS.

Clearly, I'm going to have to upgrade my UPS units, if I want to continue SKL overclocking. They just draw too much power otherwise.

 

[crashtech]

Just pre-ordered an i7-6700 from Amazon for $314.99. Guess we'll see if the BCLK trick works with it in a couple weeks or so.

 

[VirtualLarry]

Ok, I had the brilliant idea, of simply plugging the PC into the "Surge Only" outlets, instead of the "Surge+Battery" outlet instead. LOL. That way I don't overload my UPS, and I don't have to wait until next month to get a new UPS with more wattage.

So, I'm at 4.501Ghz, 136.4 BCLK, 1.3800v (BIOS), 1.344v (CPU-Z, under full load). Temps are 76-77C, according to HWMonitor. That setting is OCCT stable for 10 minutes, so I'm going to go with it to crunch with.

Unfortunately, "Package Power" isn't visible with HWMonitor when BCLK OCed, apparently. At least, it didn't show up any more.

I was going to just push it to 1.400v in BIOS, and see how high a freq. I could reach, but the UEFI showed 1.4 in RED, and warned on the side that the setting might damage things. So I settled on 1.38v. Which is still fairly high. I may not leave this as my daily-driver, I may go back to 1.200v and 4.125Ghz.

Edit: 4.501Ghz @ 1.3800v BIOS, gave me a STOP 0x101 error a few minutes into crunching with BOINC. And 4.290Ghz @ 1.250v BIOS gave me a hard-freeze. (I had been stable at that speed at 1.300v BIOS, I was trying something lower.)

So now I'm back down to a rather pedestrian 4.125Ghz @ 1.200v BIOS. At least I know that works fine, and doesn't put too much stress on the system.