How much can you trust BIOS monitor or "reported" voltages?

[BonzaiDuck]

With older, 65nm processors, given the experience of others, I could set the CPU_VTT termination voltage to between 1.4 and 1.45V, but I didn't pay as much attention to "reported" voltage values.

Now, I've set the CPU_VTT voltage to 1.45V, but watched in increments to see how the reported or monitored value changed from the reported value under the "Auto" setting.

Reviewing Graysky's OC Guide, and noting the following article here at Anandtech:

Killing a QX9650 with high CPU_VTT settings

I felt pretty smug with "carefully" notching up the VTT voltage to 1.45V, when the changes and values shown in the BIOS monitor left it at 1.39V.

Anyone have any comments about this? The BIOS version for this 680i motherboard is a "sunset-of-the-lifecyle" revision, clearing up some Penryn-Wolfdale features that needed to be addressed. Early in the board's history, there were measured discrepancies in "set," "monitored" and "real measured" voltages --- and they'd been addressed as far as I can tell with BIOS revisions through last year.

 

[Gillbot]

i never fully trust anything that monitors, if I want a sure reading I break out the voltmeter.

 

[BonzaiDuck]

I have a couple of those . . . . somewhere . . . . one digital, the other analog.

I should probably stop being lazy.

Looking more closely at the warning in the Anandtech blog, article -- what you want to call it -- it refers to a voltage spec 2.2 section in the processor specs, quoting an absolute maximum of 1.45V VTT beyond which . . . . . there will be trouble. So? My set value is 1.45V, but the reported value is 1.39V.

It may be brinksmanship, but I'm probably safe.

This E8400 is "not the best of the lot," and our fellow member nastymatt has resigned himself to a modest OC to about 3.45Ghz with it. I'll probably safely squeeze out 3.6. I had it stable at 3.82Ghz, but I didn't like the VCORE as a matter of my own "red-line" limit.

The current setting just below 1.32V yields a reported idle value of about 1.29V.

Since my "main machine" is a Q6600 with the same motherboard, and this is an "experimental flirtation" with VISTA64, I should probably loosen my caution a bit with a plan to pick up one of the E0 steppings of the E8600. With that in mind, I guess there's nothing wrong with pushing this one a little bit -- a leetle beet.

It's not as though I'd follow the example of my redneck hippie Honda mechanic. He had this '79 CIVIC 1200, he'd added a Holley carburetor (i think that's what it was), Accord 5-speed, Accord Mag wheels and brakes, Stroker crankshaft and valve-stems. I took a ride in it once -- it was like . . .. something out of Star-Trek on warp-speed -- in an empty parking lot, he'd just start from a dead stop and . . . . you wondered how you got to the other side of the lot a quarter mile away, when your wrist-watch looked as though it had stopped when he hit the accelerator. Later -- guy was a druggie I found out -- high on coke -- I think he ran the thing into a cement highway divider at over a 100 mph. And . . . . HE's dea-yud. Not "dead on arrival" __ just . . . . . instantaneously dea-yud like SPAM in a can.

"Man's got to know his limitations . . . . " Got that right, Clint . . . .