Problem Changing vcore on 4P 6166HE

[tkam]

Some I'm trying to following the steps shown here to up the vcore on my set of 6166HEs: http://hardforum.com/showpost.php?p=1038876837&postcount=552

Setting the volts seems to work fine, and switching to pstate 2 also seems to work fine but when I try to switch back to pstate 0 I get this:

wrmsr: CPU 0 cannot set MSR C0010062 to 0 2
K10Processor.cpp::forcePState - unable to write MSR

Any ideas?

 

[musky]

Which TPC version are you using? Reading through this thread, it looks like there is a bug with 0.43 related to this for MC chips.

 

[tear]

If you haven't already done so, determine limits of your CPUs. Each CPU may have different limit.
Run TurionPowerControl -l and look for Maximum Allowed VID (for every node pair); in a nutshell:

TurionPowerControl -l | egrep -- '---|Maximum.*VID'

.
Make sure _not_ to exceed maximal voltage when you perform TurionPowerControl -set ...

As musky explains, all TPC releases (as of today) suffer from a bug that may cause various
operations (not only -fo) to fail "randomly" (on some systems).

Until new release is pushed through the door, please download and install my local drop:
http://darkswarm.org/tpc/tpc-svn64-tear4.tar.gz

 

[tkam]

Thanks tear - trying out your tpc version now. All chips have the same max vcore and I'm setting it well below that.

I do think I have another problem though - seems to be anytime I set the refclock above 225 I get this error in the bios during boot:

Node0: MCA NB Error

I can hit F1 to continue but something tells me that's not a good error - and probably explains the high ht-retries count I get every time I see that error.


Edit: is there a command I can run in tpc to verify what state and vcore it's currently running at?

 

[tear]

Yeah, you may be right.

Does 225 work fine though? (is it w/o voltage bump?)
You may also (very long shot) wish to try 231 and see if that makes a difference (vide: http://hardforum.com/showpost.php?p=1038935401)

Nature of NB machine checks errors isn't really known. OCNG currently retains nominal NB freq
(1800) so, in your case, it basically follows CPU frequency. Some research has been done into
lowering nominal NB freq (to 1600) but I don't have anything that's test-ready...

 

[Patriot]

list what you are running verbatim please...
he isn't showing the first 2 steps in that post.

sudo modprobe cpuid
sudo modprobe msr

I am assuming you did this but I don't remember what errors come up if you don't.

 

[tkam]

I actually have those modules getting loaded at boot. Running the commands with tear's tpc resulted in no errors. But the higher vcore appears to not help w/ the ht-retries.

It is running stable at 225 on stock vcore and cpu temps are low even under full load only two nodes even hit 40C the rest are in the 37-39C range

 

[Core32]

So what vcore are you trying?
To run the higher OCs I have to use the maximum vcore of the "lowest maximum vcore" CPU.
Bumping up a little did not help much.
I am running 100% stable now at 258 OC with 6166HEs but also ran very stable at an OC of 262.

 

[Core32]

Edit: is there a command I can run in tpc to verify what state and vcore it's currently running at?

sudo tpc -l | grep "pstate 0"

 

[402blownstroker]

With my 6166HEs, I have had better luck running an OC value that is on the edge of the windows defined here:

OCNG BIOS Thread

 

[theGryphon]

Some I'm trying to following the steps shown here to up the vcore on my set of 6166HEs: http://hardforum.com/showpost.php?p=1038876837&postcount=552

Setting the volts seems to work fine, and switching to pstate 2 also seems to work fine but when I try to switch back to pstate 0 I get this:

wrmsr: CPU 0 cannot set MSR C0010062 to 0 2
K10Processor.cpp::forcePState - unable to write MSR

Any ideas?

The same thing happened with me, and the reason was the TPC version. With 61XX's, I never had any luck with anything but 0.41.

 

[tkam]

So what vcore are you trying?
To run the higher OCs I have to use the maximum vcore of the "lowest maximum vcore" CPU.
Bumping up a little did not help much.
I am running 100% stable now at 258 OC with 6166HEs but also ran very stable at an OC of 262.

This seemingly ended up being the issue I finally had some time today to work on this and after bumping up the max vcore to 1.05 I'm running stable at 251. Thanks for the advice!