3.14 vCore voltage... say WHAT?!

[HamNinja]

*My system and specs are at the bottom for reference.*

-------

I have been fiddling around with some overclocking and when I finished I reverted everything back in the BIOS to AUTO default settings.

I start back up and just to set a benchmark I start running Prime95 without Turbo multiplier un-locker active and no other adjustments on (as far as I know).

I get to running Prime95 for about 10 minutes with the following average readings:

MB Temp: 24c

CPU Temp: 48c

Vcore: 1.25v

All 6 cores are active and the base frequency is up slightly from 200mhz to 200.9mhz for a running frequency of about 3.214ghz.

Everything seems to be running nice and the temperature seems appropriate for the max load with stock cooler... until I suddenly get a warning from CPU-Z that says the' vCore is too high', and when I look at the monitor (part of the ASUS Probe II program) it reads 3.14v for about a second before dropping back down to about 1.25.

It did this twice in the last 30 minutes of running Prime95... as well as once dropping to about 0.50v (estimate since it went away before I could get a good look at it).

Any theories?... and thanks in advance.

-Ham Ninja

----------------------------------

MB: ASUS M4A88TD-V EVO

CPU: AMD Phenom II 1090T 3.2Ghz (stock heatsink unit)

PSU: CoolerMaster Extreme Power Plus 700w

HDD: Maxtor 7200rpm 320GB SATA-II

Windows 7 Pro

 

[HamNinja]

I forgot to mention the RAM is Corsair DDR3 2x4G 1332mhz XMS3

 

[alienware]

Don't run Prime 95. Run LinX. Prime 95 stresses your system so hard that you could have had a spike from one of your board's VRMs (voltage regulator modules).

The last time (and I mean, last time) I ran prime was on a Phenom 9950. I could smell burning and when I put my hand down the back one of the capacitors behind the CPU took the skin off my thumb. Honestly mate, there is no reason on this earth why you need to send your system to hell to check stability. The first test to run is LinX. Once that passes just go about your daily routine. If you get a bsod or a reboot adjust things accordingly but don't make your system convulse just to see if it's stable.

 

[HamNinja]

I knew it was hard.... But not that bad, thanks For the notice. You may have just saved me!

 

[alienware]

It's the same thing with Furmark and, the reason why the GPU manus have put throttling in place.

Furmark really does serve no purpose whatsoever other than battering the absolute hell out of your card. I mean sure, it can find instability but then so can playing a game

Prime and furmark are almost like a sado's dream. Why people want to shorten the life of their hardware making it do things it will likely never ever do is beyond me tbh.

 

[HamNinja]

Not really sure how to run the LinX program... what problem size, what ram to allocate, etc.

 

[tinytomlogan]

Just sounds like a busted voltage proba mate. Mr Strawberrys UD& says its running the CPU at 2v Constantly its not though.

If it was at 3v your temps would rocket instantly even at idle.

 

[alienware]

Not really sure how to run the LinX program... what problem size, what ram to allocate, etc.

Leave everything at default and run at least 10 passes. It's not your ram you need to test. This is why when I did use Prime I always went with the small FFT. No point in battering the poo out of your memory too unless you suspect a fault with it

Tom could well be right there tho, good call. I forgot about Strawberry's issue.

 

[HamNinja]

Well it seems whenever I put the CPU to the test in LinX the vCore seems to hover between 1.5v and 1.6v while holding around 55c. If it is due to a bad sensor probe.. can it lead to damage to the CPU? The board is under warranty so I should be able to pack it up and go swap it out.

 

[tinytomlogan]

have you got the Vcore set to auto by any chance?

 

[HamNinja]

I tried it with auto/manual set(1.3v)... both when OC'd to 3.6 and at stock 3.2. All set in BIOS, as well as making sure the ASUS Turbo feature was deactivated. The 1.5v+ read happens in all 4 combo's. It doesn't seem to do any harm and the temp doesn't go above 55c.

I have had it stable at ~201mhz/18.5-multiplier [3.6GHZ] with the vCore between 1.4-1.57v and 46c for some time without crash, error, or issue. So if this is normal - I'll leave it at that with the hope that there is some sort of protection in case things go out of hand. The board I can swap out easy... the CPU is a different story.

---Thanks all for the input!

 

[HamNinja]

FOLLOWUP

So far things seem to be more stable now. vCore still fluctuates between 1.25-1.55 when running LinX, but everything seems to run well at 3.6ghz with a temp of about 50c with stock cooling. Once I get the Corsair H50 and cool (pun intneded) case I'm saving up for I'll see if I can push it to 4ghz.

The latest LinX test has me running around 60-Gflops and for now that seems to be my max. When I try to push it to 65Gflops things crash.

 

[Ov3erclock3R]

Don't run Prime 95. Run LinX. Prime 95 stresses your system so hard that you could have had a spike from one of your board's VRMs (voltage regulator modules).

The last time (and I mean, last time) I ran prime was on a Phenom 9950. I could smell burning and when I put my hand down the back one of the capacitors behind the CPU took the skin off my thumb. Honestly mate, there is no reason on this earth why you need to send your system to hell to check stability. The first test to run is LinX. Once that passes just go about your daily routine. If you get a bsod or a reboot adjust things accordingly but don't make your system convulse just to see if it's stable.

Im glad you said that i have run quite a few stress tests using prime and to know its not worth using for that reason i deleted it, i tried dl'ing LinX but could find anywhere dl it on google...what are your thoughts on Burn in bud ?

Edit: just found it.

I have just run LinX and my temps are 80, 77 , 74, 74 at max 100% load, when i run prime 95 i only go like low late 60's low 70's luckly i have a noctua for cooling.

@HamNinja i only got like 51 Gflops and i see you got highs of 59/60 why is that, and a what does it test, its a newbie question, cos OC'ing isnt really my strong point at the moment im just learning.

 

[Tony]

10 passes is not enough, I failed the linx run at 17. I always aim for 20 quick and overnight for ensure stability.

P95 doesn't stress your cpu more. Linx does. You can tell by the temps.

@ov3erclock3r

It's not 100% stable imo. I could be 'stable' at 4.4ghz but produce less gflops then i would if i'm at 4.0 which doesn't make any sense.

 

[HamNinja]

3.6ghz seemed to be, at that point, the most stable setting I could get it to under automatic settings for voltages and RAM, when I tried to get it higher the GFLOPS dropped and stability was iffy at best. I just got a Corsair H50 cooler and will be purchasing a new case tomorrow... so after putting everything back together I will try pushing it a little further and adjust voltages with 4ghz as my goal.

I'm going to do some more research in the meantime since apparently I've got to get the RAM frequency/voltage to match what I'm running the CPU at in order for the thing to even work. I'm still concerned about the over-voltages since it seems to fluctuate even when I manually set the vCore and turn the auto-tuning stuff off.

As soon as I find some settings that work I'll share.

(on a related note) I found it interesting that the built-in auto-overclocking feature in the BIOS focuses on increasing the base frequency really high (236)... but didn't seem to be stable at that level. However, setting it lower and focusing on the multiplier seemed to work better. I'm still a newbie at all this, but this whole thing is proving to be rather fascinating and challenging.

 

[I Hunta x]

A little fluctuation isnt unheard of on set volts, ive seen a few systems go up or down around .1 volt depending how the load changes ect. But your best off getting it off of auto voltages asap, it will end up being better for your hardware tbh as you know it will be in the safe zones and not spiking to god knows what.

Im not 100% sure on safe volts for the current amds but if i remember right its around the 1.55v mark? would need someone else to give you a bit of a heads up on that area.

 

[mezmir]

Don't run Prime 95. Run LinX. Prime 95 stresses your system so hard that you could have had a spike from one of your board's VRMs (voltage regulator modules).

The last time (and I mean, last time) I ran prime was on a Phenom 9950. I could smell burning and when I put my hand down the back one of the capacitors behind the CPU took the skin off my thumb. Honestly mate, there is no reason on this earth why you need to send your system to hell to check stability. The first test to run is LinX. Once that passes just go about your daily routine. If you get a bsod or a reboot adjust things accordingly but don't make your system convulse just to see if it's stable.

That's scary hot :/