Skylake hard locks seeminly when IDLE / CStates ON

[GGlyn]

Myself and others have been struggling with a different hard lock problem with Skylake and we have ran out of options but to assume it is a CPU fault/issue. Does anyone have any comments or feedback?

The long thread is over here, but I will summarise. http://www.tomshardware.co.uk/forum/id-2830772/skylake-build-randomly-freezing-crashing/page-7.html New Skylake Build Randomly Freezing/Crashing - Page 7 - windows 10 - Windows 10

Our systems hard lock randomly but appears on the most part to be when IDLE or next to IDLE. The hardlock can come when playing video, browsing (reading) or mostly when not interacting at all. Some (like myself) noted that they seem to occur more often if you are doing a large amount of disk IO (partition move for example or copying 40+ GB of data disk to disk).

Hard locks happen in either Windows (7,8,10) or Linux.

With CStates on, the crashes are often and usually within 10 minutes of IDLE. With CStates disabled completely most find that the system remains "mainly" stable for days. Generally, no one has reported locks when gaming or under high CPU load (so I do not think this is the Prime95 example but then, since compilers / drivers use the latest instructions (according to CPUID) I cant be sure).

The symptoms happen on many different manufacturers motherboards, RAM types and setups. We have successfully removed or replaced all components and the only change that seems to make a difference is CPU for a few. CStates off in the BIOS is absolutely a guarenteed way to stabilize significantly but not fix.

I changed CPU (after trying everything else) and find it now mostly stable with CStates ON. I must be clear that my system is 100% identical, CPU # 1 crashes all the time and must have CStates off, CPU # 2 does not crash so often even with CStates ON. One may argue this is a problem with my PC, a fault, but with so many people having the exact same issues I cant believe it is a defect. And with Cstates/SpeedStep so specifically (a known new part of Skylakes architecture), I remain convinced that this is something real.

Others were not so lucky, a replacement CPU did not make the situation better. Batches from different factories as well etc etc

My new CPU has hard locked but not when IDLE like the last one (not yet anyway). Mine can lock when under full load, video transcoding, but not more than 1 in 10 sessions. Cstates ON seems equally stable on this CPU as OFF.

Thanks for any suggestions you can give.

[MRuni]

You have a different issue.

Issue still unresolved.

*Update: I have reduced the amount of idle freezing by the following

1. Turn on XMP Profile.

2. Turn off C8

3. Overclock CPU to 4.6 Ghz*dont touch voltage

Save profile. I have only froze 2x in the past 7 days. Massive change from the 2 to 5 times a day lock up when idle.

[GGlyn]

For reference

CSTATES

With my original processor (i7 6700K) it was IMPOSSIBLE to resolve the IDLE locks without disabling CStates (all of them). With CStates enabled, the machine would hang within a few minutes to one hour. With CStates disabled, the machine was IDLE stable as far as I could test in the time frame I had that CPU.

With the new processor I can have CStates enabled without problem, there are NO IDLE issues after weeks of confirmation testing. No other component was changed, and all others had been ruled out though swapping/testing before this conclusion.

MEMORY

I did experience a different kind of lock (under high load) but rare. This error was there before the CPU swap but was hard to diagnose in addition to the CStates. Basically when transcoding and/or copying large files I would hard lock too, "sometimes". This second "symptom" was causing confusion with the first, and with the CState stability solved, this was more clear.

I assumed that the Prime95 error was the issue, until that fix did nothing.

I ran HCI memtest on the memory (with its default settings 2666/1.2v) and had errors at around the 300% mark. Memtest86 PASSED so please DO NOT use it for confirmation of memory stability. If I put the speed to JEDEC 2133Mhz and correct timings I still has errors. I pumped up the voltage to 1.25v and then it reaches 1300% of testing and this is considered stable. Note 1.25v is not enough at the memories advertised 2666Mhz (I have still to find the right voltage for its advertised frequency, but for now I am stable so wont spend more time).

So in summary. A faulty CPU caused CState issues, RMAing the CPU solved it. Still memory issues when HIGH memory load (large buffers, transcoding, copying etc) with 1.2v regardless of factory settings (SPD @2666Mhz) or standard 2133 timings. XMP created much more errors on testing with standard volts.

Pushing up to 1.25v on 2133Mhz resolved that instability. I guess I could go to 2666Mhz if I bump the RAM up to 1.30-1.35v (have not tested).

A note on HCI Memtest. Not the most friendly program in the world, and a little weird. Turn off Windows paging (virtual memory) you dont need it anyway Then for every CPU core / thread ("X") (this is 8 on i7 with hyperthreading, 4 on i5) run an instance of HCI. Now take free memory (task manager memory page will tell you), divide by the number of cores/threads ("X") and enter this into each instance of HCI. Click start on each... go to work.. sleep.. work.. sleep.. OK, check If you have any errors, any at all, you have an issue. If you get to 1300% on all instances and no errors, your stable.

[KTran10]

You have defect CPU and just request new CPU from Intel. That is my 2c. Intel lately has issue with quality.

[DMas]

Hi,

I was following post on tomshardware.com that led me here. I'm quite sure that i have the same problem (random idle freezing, turning c-states off makes it somewhat better). There was information that this is the thread where Intel will post possible solution, so to chip in my system specs in hope it will help.

CPU: Intel i5 6500

MBO: Gigabyte GA-Z170X-Gaming 5-EU

RAM: Corsair CMK16GX4M2A2666C16

GPU: Gigabyte GV-N960G1 GAMING-4GD in SLI

PSU: Corsair CS750M

SSD: Samsung 840EVO 250GB

Since computer is now stable enough to play after work I won't be sending back CPU or MBO just yet. I bought them from two different shops and I have a feeling that they'll try to blame each other, so I'm hoping for some confirmation from Intel that this is fixable by microcode and/or BIOS update, or whatever official.

Cheers!

[DMas]

Few days ago I installed latest F5 version of bios for my Gigabyte GA-Z170X-Gaming 5-EU. Computer ran 48h without freezes, so I guess they did something.

[FParr]

Perhaps the failure of all this is the WDDM 2.0 of Windows 10. Look at this:

You know if there is any product of Microsoft to leave feedback ?, If any member of Intel insurance could confirm what we would go.

[GGlyn]

FYI, the CSTATES issue also happens in Linux. Ubuntu 14.04, 15.10 tested.

If this was an OS issues, RMA of the CPU would not solve it. But it did.

[FParr]

Thanks, interesting, these operating systems had not tried it.

The funny thing is as I commented in another post, is that with Windows 7 64bit works great.Why?

Correct me if I'm wrong, but as you indicate on April 14, was not Skylake market.

It would be a good news that the latest fabrications Skylake longer work without failure freezing.

I bought the i5 6600k in November, did RMA in December and the last on January 10 and still does not work.

Skokie, Can you tell me what date the RMA have made your processor you freezing fault disappears, and so have an orientation to make me too? .And If you do not mind, how many hours you failed and how many have you kept working to make sure ?.

[GGlyn]

According to someone here, Windows 7 does not utilize power saving in the same way as Windows 10/linux. It is less likely for a core to become fully IDLE under Windows 7, so the CState behaviour will be different (specifically with regards to FET low voltage behaviour).

Although drivers and OS can have an effect, and although BIOS and platform are responsible for CStates, it is clear that this IDLE Cstate problem is unrelated to a specific driver/BIOS issue and more related to poor tolerances on the CPU/Platform. The FET explanation is the most likely to my mind.

Changing a CPU can resolve it, but others have found it does not. This suggests that the problem is wider than initially thought and a little Pot Luck. As far as I know it is only the K series that suffers it (unlocked).

My replacement CPU was made in the same factory but before my original. Others report problems remain with their new CPUs and those were from a different factory.

There is the memory issue which develops symptoms under load. It may be that this is causing confusion in the CStates IDLE diagnosis. But they are different things.

[FParr]

Thanks for your quick response.

I bought it in a store, not directly on Intel, so may rolls processors flawlessly as in the factory have not arrived.

Now I understand a little better what of that works on Windows 7 and the other not.

Maybe try a third time lucky and make a RMA, before it is too late and the warranty expires. And I will comment.

When did your Last RMA?

[MFlet2]

As far as I know it is only the K series that suffers it (unlocked).

6500 here, no K. I have the same issues.

Disabling C-State does not helping. I tried a different motherboard to no avail.

[GGlyn]

I suggest if disabling C-States does not help, then you have a different problem to this.

I would run HCI Memtest and confirm it is not your memory causing hardlocks. I guess it is likely memory.

[MFlet2]

I got to 450%+ testing my consair vengeance lpx ram before a lock. Tried both sticks by themselves.

Also just went out and bought overpriced crucial ram from a local store, still getting hardlocks.

I've tried different psus (the one I'm using now on this computer specifically).

If it isn't the cpu, I don't know what else it could be.

[GGlyn]

Ok, this is a shame.

HCI Memtest would not directly hard lock your computer since the memory allocated is simply used for read and write (nothing critical) and hence would not hard lock directly due to its activities.

If CStates dont help you then I would still focus on the Memory. I assume the locks happen usually when gaming or copying files or such? If yes, then probably your memory is not compatible. You can check against the Motherboard site but be aware, use the actual model number, not the simplified brand/size (So dont assume all corsair 4GBx2 are the same.. they are not.. find the actual model number for your memory and reference the compatibility list).

Your problem sounds severe enough that it should be simple to trace. Check your timings (some BIOS set it incorrectly on Auto). Check your IO voltage (should 1.00v and not 0.9v like some BIOSes set it by auto). Use 2133Mhz frequency (and JEDEC timings) for your RAM, regardless if it is 2666Mhz or some other.

Try it then, if its better but not reaching over 1000% on HCI memtest, then up the voltage a little 1.25v for DRAM Vcc.

In the end, if all that doesn't get you any feeling of "improvement", you should consider next steps.

[ctroy]

Same issue here

PC is 100% stable with C-states disabled, but will hard lock if they are on

MOBO: Asus Z170 deluxe

CPU: i7 6700k @ stock

RAM: Gskill Ripjaws V 3200

PSU: Corsair HX850W

GFX: GTX 980 ti

OS: Windows 10 pro 64

[FParr]

I disconnected the C-State, and continues to fail freezing, other than the processor temperature becomes very unstable with ups and very sudden rises from 3º to 6º.

MB: Asus Maximus Hero VIII

CPU: i5 6600K - 3,9Ghz

RAM: GSkill TridentZ 3200 Mhz

PSU: Tacens Mars Gaming 750 Silver

GFX: GTX 970

OS: Windows 10 pro 64 UEFI

[dmone2]

I know most of you here are convinced that this is a Skylake c-state issue, and it may be. But just in case, if any of you are experiencing drive read spikes coincident with the cpu locks, try updating your SATA driver. My freezing problems completely disappeared with a simple SATA driver update. Use Slim Drivers freeware.

The chances that Intel released an entire load of chips with bad architecture are next to nil. As evidenced here and in other forums, RMA is likely not the solution. If you are not finding a solution by playing with voltage/power, look elsewhere. Also, the fact that Intel has released no official statement other than the big Prime95 issue should indicate that your problem lies in another area.

Don't judge me, I'm a brand new noob.

[FParr]

Hello dopemoney81 and for all, I have thought about doing RMA processor which comes on the recommendation of some users here as a possible solution week.

I saw your recommendation to use Slim Drivers freeware, but not if in my case will make effect and use only one SATA port for the DVD drive, and I have a PCIe SSD M.2 it.

Sorry for the translation of Google

Cheers.

[JCort5]

How did you go about loading the configurations from the BIOS? I have loaded them on a flash drive and went to BIOS but wasn't sure where to go from there. I'm a bit of a noob on this as well.

[dmone2]

in your bios menu you should see an option to update bios, and then you should see option to update bios from flash media or usb, make sure folders are named appropriately for bios update to work, directions can be found on gigabyte website

[JCort5]

FIXED MY PROBLEM!!!!

Ok. I seemed to found a fix to my problem. Looking on another forum it was suggested to check your motherboard website for a list of compatible memory. I opened my case to check my memory as this is a new build from newegg.com It turns out that the RAM they put in my CPU is not compatible with my memory. How could a CPU builder overlook such a simple flaw? Contemplating calling them so they can send me a whole new set of COMPATIBLE RAM! In any rate I upped my timings to coincide with xmp rating without changing the mhz to the xmp rating as that seemed to prevent booting. Upping the timings to the suggested timings manually has fixed my freezing.

Here are my details of my CPU uptime being stable for over 2 days now when before it would freeze after 10 minutes of being idle. I still have C states off as I don't want to mess up a bad thing.

CPU

Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz

Maximum speed: 4.00 GHz

Sockets: 1

Cores: 4

Logical processors: 8

Virtualization: Enabled

L1 cache: 256 KB

L2 cache: 1.0 MB

L3 cache: 8.0 MB

Utilization 0%

Speed 0.76 GHz

Up time 2:06:35:04

Processes 61

Threads 1454

Handles 33913

I hope this helps someone out there!