6700K - IA: Max Turbo Limit + IA: Turbo Attenuation (MCT) flags on hwinfo64

This is roughly the build: http://pcpartpicker.com/p/dBCnf7
The rams are Corsair Dominator DDR4's instead, it's my brothers system, on mine I replaced the RAMs with TridentZ as my pair of Corsair's were faulty

The CPU runs at 1.23V, not Auto, everything else is default

My brother told me that he experienced a momentary freeze/glitch on Black Ops III, I told him to run a quick memtest after this happens and keep hwinfo64 open at all times

When he opened hwinfo64, even though there was no load on this system, these 2 flags got set at one point:

IA: Max Turbo Limit
IA: Turbo Attenuation (MCT)

I'm guessing it might be insufficient voltage or something, any ideas?

Restarting the system doesn't reset the flag, the CPU also prefers to run with 36/38x multiplier, at Prime95, the multiplier is 36

Set the voltage back to Auto, didn't change anything, before using a constant 1.23V voltage, Prime95 used to push 1.4V's at Auto voltages, currently, with this flag in place, the Auto voltages are also at 1.2's

So somehow this flag got set, it's persistent and it seems to limit the Turbo behaviour

I've only seen this online that is remotely related: https://communities.intel.com/thread/92438

This is 1.2 BIOS, I'm using 1.34 Beta myself, haven't cleared the CMOS or moved this system to the 1.34 Beta yet

Is there any issue? Are you not turboing to where you are supposed to? Check that first...

I wouldn't hang my hat on BLOPS 3 hitching once to be a sign of an issue.. that game sucks down ram like a man that hasn't drank water in days...

Will check the issue further, but the initial login was also taking like 10 seconds with 1.2V, with Auto it became instantaneous like it should have been, might be unrelated (the windows 10 login // boot in general)

Also on Prime95 it seemed like the CPU was throttling, should check the Cinebench score later on to see whether the score is reduced

I also saw that there was a 1.3 BIOS on MSI Global, it makes sense to flash that first

Yeah, id confirm you have an issue first.

As an update, the flags persisted through a BIOS update and CMOS reset, Cinebench score is 878, my system (identical, second system) gets 910ish, has different RAM's tho

I remember this specific system getting 900 with Adaptive Voltage too, currently the Adaptive voltage goes easy on the system, very easy, so the flags definitely changed the behaviour a bit, but can't verify it 100%

Using Intel Extreme Tuning Utility I couldn't see anything out of the ordinary, the Turbo was reported 42x for 1 core and 40x for 2,3,4 cores, the benchmark score there was 1221, which is similar to my system

Still it's bizarre that the CPU goes on with these flags, I wonder whether it's CPU's own way of making a faulty CPU work

Edit: Actually, upon closer comparison, the flags seem to disable 40X+ Turbo on multiple cores, on my system 1,2,3,4 cores can all turbo to 42X, this system is limited by 40X on multiple cores

So it's <5% performance loss + paranoia

kaans said:

As an update, the flags persisted through a BIOS update and CMOS reset, Cinebench score is 878, my system (identical, second system) gets 910ish, has different RAM's tho

I remember this specific system getting 900 with Adaptive Voltage too, currently the Adaptive voltage goes easy on the system, very easy, so the flags definitely changed the behaviour a bit, but can't verify it 100%

Using Intel Extreme Tuning Utility I couldn't see anything out of the ordinary, the Turbo was reported 42x for 1 core and 40x for 2,3,4 cores, the benchmark score there was 1221, which is similar to my system

Still it's bizarre that the CPU goes on with these flags, I wonder whether it's CPU's own way of making a faulty CPU work

Click to expand...

I don't see a problem...

Updated the post

The multi-core Turbo capability seems to be disabled

So that system never ramps up to 4.2GHz?

Nope

Cores 2,3,4 are capped at 40x

My theory: Insufficient voltage during a previous Turbo session activated the flag, this CPU might have been a power hungry one, 1.23V might be insufficient

Yet it doesn't explain why the flag is persistent and whether it's there for life

Also from the data I've seen, 1.23V should have been enough for non-OC/regular usage

1.23v is very low for its VID/Stock voltage...

Isn't 6700/6700K 1.2V at stock?

I've tried my own CPU with 1.2V, it performed well, but I decided to use both of them with 1.23V, 0.03V to add some room

So far my own CPU performs well at 1.23V, I used to see the Voltage bumping to 1.4V's on Adaptive, the temps reaching 70C+'s, that's where the decision to run it at 1.23V comes from, it's cooler, the performance is the same, the stress tests didn't produce issues

According to these findings: http://overclocking.guide/skylake-overclocking-power-consumption-and-voltage-scaling/ - the CPU should be capable of 4.4GHZ at 1.23V, 4.2GHZ at 1.16, so 1.23V seemed more than enough

Obviously every CPU has different needs, also turning back the Adaptive Voltage on didn't prevent the issue, so it might even be unrelated

As I said, 1.2v is really low for VID/Stock voltage. I haven't seen one lower than 1.25v personally. I have had two, one was 1.3125 and the other 1.2975. That guide manually lowered the voltage. If yours works that low, then great! But it is below stock I would imagine.


If you would like to know the stock voltage of your CPU. Set the BIOS back to factory defaults, and boot back into it... then check what the core voltage is at (in the BIOS).

Another guide.. their chip starts at 1.3v as well: http://www.corsair.com/en-us/blog/2015/august/overclocking_i7-6700k

Yes but those are the Auto voltages that the Motherboard pushes, that guide seems more like an observation, BIOS'es are not really optimised at this point IMO, I locked the voltages to prevent the 1.4V spikes that were terrifying me

http://www.anandtech.com/show/9533/intel-i7-6700k-overclocking-4-8-ghz on all the OC reviews I've seen, 1.2V was more than enough for stock 40x/42x

Tho when you leave motherboard alone, the voltage was 1.26V on one system, 1.29V on the other one, even in BIOS, seemed too much

(At the time my system was unstable, due to a faulty RAM and because XMP never worked, even after buying new RAM, that's why I lost trust in BIOS, the Auto voltages seemed too funky, funky enough to burn the CPU)

Yes. We say not to OVERCLOCK with using auto voltage. However at stock it is what it is. Most all chips are set higher than they can go and can lower at stock. Not to mention, software readings in windows are mostly inaccurate.

BIOS are maturing, yes, but we haven't seen stock voltage of the CPU drop as they mature. You are barking up the wrong tree me thinks.

Lol, maybe

The voltage dynamics change tho, I've previously saw 1.4V's, but with updated BIOS's I haven't seen anything close yet

These are kind of off topic tho, as the issue is persistent

It's interesting that there is not one report of "IA: Turbo Attenuation" online, I'm guessing no one paid attention, or it might be a unique issue

I should probably ask here: https://communities.intel.com/community/tech/processors

Intel seems to reply to issues occasionally

Im guessing you are reading software voltages and not taking readings from a multimeter?

Yes, no multimeter readings, I'm checking them from BIOS and hwinfo64 or similar programs, the readings are consistent

As an update, disabling Turbo soothes the situation, the CPU/mobo is no longer capable of Turbo it seems ...

(The flags are off when Turbo is off, so they aren't persistent, but rather the setup was unstable with Turbo)

I bumped the voltage up to 1.24V just in case, performed well with Cinebench 870+ / Intel Benchmark 1240+

Weird. Disable turbo and raise multiplier to 42x... Done.

Currently at 40x, staying away from any kind of OC until the BIOS waters settle a bit