768k Skylake Problem/Bug

[s1riker]

Originally Posted by chalsall

Are you able to replicate this on any other machines?

Unfortunately, don't have any other systems to test on.

[s1riker]

Originally Posted by pegnose

Have you read what I wrote above? Probably you have, but I just post it again:

From what I have heard, hard lock during idle could be a third problem. I had this when I had ASPM enabled for the link between SA and the PCH. Others suggested the PSU being incompatible with Haswell C-States (6/7). You could try disabling all power saving options altogether.

Did you do that? Also in Windows?

I tried two recent, relative high end power supplies, and both exhibited the exact same issue. The next thing I was going to try was to disable C-States one at a time and see what happens. Although, I was hesitant to go down that route because I'm not fond of having the system waste more power than it needs to, given that I leave it running 24/7 but for a couple weeks, I'll give it a shot.

Also, to the above poster who suggested I try 1 stick of RAM. Did that, still hung, but thanks for the suggestion.

[pegnose]

Originally Posted by Madpoo

By the way for everyone following along...

The www.mersenne.org server has seen a decent little boost in traffic to the download page, mostly from links on PC World.

I'm sure many of those are people interested in testing their own Skylake to see if it exhibits the issue, but maybe some will stick around and test a few exponents.

Any particular instructions on that? Program version, settings?

[pegnose]

Originally Posted by s1riker

I tried two recent, relative high end power supplies, and both exhibited the exact same issue. The next thing I was going to try was to disable C-States one at a time and see what happens. Although, I was hesitant to go down that route because I'm not fond of having the system waste more power than it needs to, given that I leave it running 24/7 but for a couple weeks, I'll give it a shot.

Also, to the above poster who suggested I try 1 stick of RAM. Did that, still hung, but thanks for the suggestion.

Make sure, the PSU vendor explicitly states compatibility with Haswell C-states. Then disable C-state 8 in Bios, because to my knowledge, Haswell C-states are 6/7. Doesn't have to mean anything, but I did all this, and I don't have the issue. After reading your thread at Toms I let my machine idle overnight. Was still there in the morning. I also have the microcode 39. Hyperthreading, C-states, Turbo boost... all enabled. ASUS M. VIII Hero, Bios 1302.

The second thing is ASPM (advanced state power management) for PCIe components. Make sure, everything related to this is explicitly disabled in the Bios. Also make sure PCIe power management in Windows power settings is set to "Maximum performance". OS's ability to do control ASPM is called PCIe Native Power Managment, at least in ASUS Bioses.

And finally I recommend a CMOS reset by inverting your CMOS battery in the socket (for at least 10 min) to make sure that (really, really) everything is gone.

[s1riker]

Originally Posted by pegnose

Make sure, the PSU vendor explicitly states compatibility with Haswell C-states. Then disable C-state 8 in Bios, because to my knowledge, Haswell C-states are 6/7. Doesn't have to mean anything, but I did all this, and I don't have the issue. After reading your thread at Toms I let my machine idle overnight. Was still there in the morning. I also have the microcode 39. Hyperthreading, C-states, Turbo boost... all enabled. ASUS M. VIII Hero, Bios 1302.

The second thing is ASPM (advanced state power management) for PCIe components. Make sure, everything related to this is explicitly disabled in the Bios. Also make sure PCIe power management in Windows power settings is set to "Maximum performance". OS's ability to do control ASPM is called PCIe Native Power Managment, at least in ASUS Bioses.

And finally I recommend a CMOS reset by inverting your CMOS battery in the socket (for at least 10 min) to make sure that (really, really) everything is gone.

Thank you, I will try your suggestions.

[ralleh]

I can confirm the 6A MC fixed everything on my system :)

Thanks for everyone involved over here, we made it happen!

[ralleh]

Originally Posted by pegnose

Ok, thank you. I just ask because the experienced testers here said they didn't have any trouble whatsoever with other settings then AVX FFTs with 768k. So if I take them by their word, my issue must be a different one. I was hoping that some of those could tell me that they had issues with other test constellations, but only less frequently.

I can confirm there were no issues with other tests before. I tested 800k for 12 hours+ successfully with the old MC.

Your issues sound more like a regular system instability due to bad overclock or component issues. 800k is used to verify memory overclocking stability usually (was found to be one of the best tests for RAM overclock), so you might want to try another kit or check if the ram is the problem in general (some DDR4 kits that were released for X99 systems cause severe stability problems on some skylake motherboards).

Also I had never any issues/freezes when I was using my rig for gaming, when using the old MC, too.

Originally Posted by s1riker

Although, I was hesitant to go down that route because I'm not fond of having the system waste more power than it needs to, given that I leave it running 24/7 but for a couple weeks, I'll give it a shot.

The difference in power usage is way smaller than you might think (on more recent systems). Mostly <5W on DC/Skylake systems I tested ;)

-

/Edit: In addition to my post above: The usual version of P2.00 for the ASRock Z170 OCF does not include the MC 6A yet, it was modded into the UEFI. German retailer JZ Electronics offers such modded UEFIs for his customers (and non-customers who bother to register in his forum). Here is the link to the modded ASRock UEFIs.

[pegnose]

Originally Posted by ralleh

I can confirm there were no issues with other tests before. I tested 800k for 12 hours+ successfully with the old MC.

Your issues sound more like a regular system instability due to bad overclock or component issues. 800k is used to verify memory overclocking stability usually (was found to be one of the best tests for RAM overclock), so you might want to try another kit or check if the ram is the problem in general (some DDR4 kits that were released for X99 systems cause severe stability problems on some skylake motherboards).

Also I had never any issues/freezes when I was using my rig for gaming, when using the old MC, too.

Thanks for your follow-up and concern, ralleh! It had turned out that Crucial shippes the same hardware with different timings, one of which wasn't compatible with my mobo. So a software (bios) fix did the trick (tRAS: 39 -> 40).

[pegnose]

Erm... why do I have an Alien avatar?!? Is this mod humor?

INDEED IT IS. YOU HAVE BEEN CHOSEN. CONSIDER IT A VERY SMALL BADGE OF HONOR

[Dubslow]

The various animals that (secretly?) run this place get bored from time to time...

[pegnose]

Originally Posted by Dubslow

The various animals that (secretly?) run this place get bored from time to time...

And here I thought you wanted to distance yourself by marking me as being from outer space. ;)

So, thank you... I guess. Already starting to feel minimally honored. It is slightly tickling...

Also started working on some prime factors in my idle time. Is it a problem to cut the internet connection while Prime95 is doing the hard work?