X2 Bug Warning: Thermal throttling bug

imported_Starman

Senior member
Jun 1, 2005
281
0
0
There's a bug in the current revisions of X2 CPUs wherein either
the thermal safety throttling may be unexpectedly forced off, or wherein
its use could cause the CPU core voltage to spike, possibly frying the CPU.

If the thermal throttling is forced off e.g. in the BIOS as is suggested
as a "work-around" for this bug, it's possible that some BIOSs would permit
the CPU to continue to run continually even if it has reached temperatures
beyond which the CPU could be damaged; normally thermal
STPCLK speed throttling would reduce the CPU clock to prevent this
damage from being so likely.

If thermal throttling is NOT disabled as AMD suggests the BIOS does, *OR*
if user selected ACPI ower saving / utility functions triggers STPCLK
mode while the CPU clock is running at normal speeds, the voltage spikes
caused by the instantaneous 100 Watt <--> 0 Watt Vcore load change could
cause the system to spike or undervolt Vcore leading to either a crash
or very possibly a fried CPU!

Everyone should confirm with their BIOS vendor and settings for their
ACPI / thermal monitoring options in the BIOS that they're running the core
at safe temperatures and in a way that will not trigger STPCLK throttling /
power saving!

Here's the info from AMD:

http://www.amd.com/us-en/assets/con..._docs/25759.pdf

Revision Guide for AMD Athlon? 64 and AMD Opteron? Processors 25759 3.51 06/16/05

124 STPCLK Throttling Causes Violation of VDD_ac Specification
on Some Dual-Core Processors

Description
STPCLK throttling during maximum power consumption operation generates large load steps and voltage transients that can violate the transient
voltage specification (VDD_ac). This problem is exposed on 4-layer
motherboards with single power and ground planes when the
core frequency is >2000 MHz.

Potential Effect on System
Violation of the VDD_ac specification leading to unpredictable operation.

Suggested Workaround
Do not enable STPCLK throttling on affected dual-core processors above 2000 MHz.

Fix Planned: Yes
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
This is a repost and has been discussed with me personally about 3 times in threads over the last week....

It is being blown out of proportion from what I read and actually many boards (higher quality ocer boards) may not have the issue....

No reports of these happening to date that I am aware of from american X2 owners to foreign users who have had them a bit longer, to review sites that have had them the longest...

MOst ocers run with throttling off anyways, and monitor their temps and voltages...I dont see a concern for most of us.....


Always good to post the info for the newbs though!!!
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Duvie, I wonder if it's been the cause of my system instability. I have my X2 lock up several times a day. Perhaps the voltage spikes are causing my crashes (?)
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
That is likely power supply is not good enough....what do you have for power supply???

It is always worth investigating but if this thing was going to fry chips I am sure someone would hav edone it by now....

List board, power supply, and if you are doing an ocing,please!!!

By the way MarkFW900 had same issue with his Abit AV8 which was running a 380watt antec with a 6800GT and an X2...he noticed the issue after 2.4ghz when running 3dmark...clear sign of power issue and we fixed it by having him swap out for the higher 12v amped 430 true power Antec....
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
What vid card??? How many drives?? You have the 4800+ to right??? You may be right on the cusp if you have a high end power sucking vid card and a slightly undervolting mobo!!!

Try bumping the vcore up a notch if it is not being reported or recognized at 1.35v actual??
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Sorry, I have an X800Pro, two IDE hard drives, and two optical drives. I've run into the instability using both an X2 4800+ and the 4400+ (which is what I'm currently using). In addition, I only have two 80mm case fans (but the CPU temperature at idle is in the mid-30's and under load only climbs to just above 40 degrees).
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Could very well be power supply..how are the 12v rails looking and are you seeing any large jumping with the vcore???

When exactly do the freezes occur?? random or at the launching of 3d intensive things, or when multitasking and have many things going including spinning f the drives???
 

Pabster

Lifer
Apr 15, 2001
16,986
1
0
I'm not a big fan of Enermax, frankly. I've always used Antec, although some might not approve of that either. You might look at your +12V rail and see how many amps it can handle. I like 20+ although that is debatable.

This whole "bug" doesn't appear to be a big deal.
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Originally posted by: Duvie
Try bumping the vcore up a notch if it is not being reported or recognized at 1.35v actual??

My Asus Probe Utility is reporting that the Vcore is fluctuating between 1.23-1.27 volts :confused:
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
That is waaaaayyyy too low...The old Asus V droop...it is likely you dont have 100% offcial bios or they have a bug cause you should be much higher considering default vcore is 1.35-1.4v....
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Originally posted by: Duvie
That is waaaaayyyy too low...The old Asus V droop...it is likely you dont have 100% offcial bios or they have a bug cause you should be much higher considering default vcore is 1.35-1.4v....

Yeah, and it's not like I have the BIOS automatically setting it. I've configured it manually for 1.35v according the AMD spec. Would this be a BIOS issue or a possible problem with the motherboard itself?
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
MOst boards just undervolt (not likely the power supply) Low +12v is the power supply. The mobo takes the +12v rail and reduces it down to the vcore..How well it does this is the question...If I set my MSI to 1.4v the default for a winchester I will get 1.33-1.34v actual...So it is common....

I would bump the vcore up until you get actual volts as reported by bios and all your other monitoring apps to be 1.35v idle....

Does the asus vcore go up or dwon from idle upon loading it???

 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Originally posted by: Duvie
MOst boards just undervolt (not likely the power supply) Low +12v is the power supply. The mobo takes the +12v rail and reduces it down to the vcore..How well it does this is the question...If I set my MSI to 1.4v the default for a winchester I will get 1.33-1.34v actual...So it is common....

I would bump the vcore up until you get actual volts as reported by bios and all your other monitoring apps to be 1.35v idle....

Does the asus vcore go up or dwon from idle upon loading it???

When I set my Vcore to 1.35v and reboot back into BIOS, the Vcore is shown as 1.296 (in BIOS). Booting with this setting into Windows right now and using Asus Probe to check it, the Vcore is fluctuating between these numbers: 1.23 - 1.246 - 1.294 - 1.262 - 1.214 (!)

When I tried setting the Vcore to the next step up, 1.375 - upon reboot the BIOS reported the Vcore to be 1.55v! Set the Vcore to 1.4 and reboot - BIOS reports the same 1.55v! Man this is annoying. I'm open to suggestions.
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Originally posted by: PetNorth
I guess you have installed 1013 BIOS version (first one with X2 support). Try the lattest beta BIOS, 1014.006 perhaps it resolves this undervolt vcore issue

http://support.asus.com/download/downlo...spx?SLanguage=en-us&model=A8V%20Deluxe

Actually, I had stability issues on 1013 (and it disabled 1T command rate) so I updated tp Beta 1014 and that's what I'm running now. I'm going to try and revert BACK to 1013 and see if I still have this Vcore problem.
 

ryanv12

Senior member
May 4, 2005
920
0
0
This isn't good news since I have an A8V Deluxe waiting for my X2.

Hope they release a new BIOS soon. I guess I'll stick with 1013 and have to work at 2T :\
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
So yeah, the Beta BIOS versions suck. I went back to 1013 and everything is fine. The Vcore is still fluctuating wildy, but at least what I enter into the BIOS is consistent. For instance, entering 1.35 I got just below that (always above 1.3) and entering 1.375 into the BIOS I'm getting 1.326 and 1.358 and so on. Before when entering anything higher than 1.35 I'd see the Vcore shoot up to 1.55v :thumbsdown:

...I'll just wait for more BIOS revisions, but in the meantime is this much fluctuation normal?
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I usually only seen it on flaky mobos.....My MSI neo board fluctuates like .02v at idle and loads up .02 under load and holds pretty steady....wildly jumping vcore can be a sign of a flaky power regulator on the mobo or a power supply that cant cut it...The last makes more sense if the +12v rail is really low or jumping wildly all over the place....


It is true my enermax 431 ran the system I have below minus the vid card to 2.6ghz, but I never thought it would be able to run a dual core system..hence why I spent the money to get a high end power supply I see as potentially hanging on for a few years...
 

imported_Starman

Senior member
Jun 1, 2005
281
0
0
Well according the Asus Probe my 12v rail is very steady. It just sits there delivering 12.16 and doesn't really budge at all. I manually set the Vcore to 1.375 and Probe reports it fluctuating anywhere from 1.31 to 1.39 (and everywhere in between). Mostly it's hanging out around 1.358, 1.342, 1.31...at least now it's delivering closer to what it should, but the frequent fluctuations have me a bit concerned.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Board then and I dont think any bios fix will help you there...Lets just hope it isn't really that bad....How was it with you old chip???