P5B and undervolting

[jdh79]

I have read the many threads about the Asus P5B and undervolting. If I set the VCore to auto in the BIOS, the VCORE reports as 1.12 (idle, speedstep) to 1.26 (100% load). I can specify the VCore to be 1.35 to force it to run basically at the 1.31 stock voltage. However, if I set a VCore value other than AUTO in the BIOS, SpeedStep/EIST no longer does voltage adjusments.

I have an E6700 and don't plan on overclocking, so I would prefer to leave SpeedStep on to save power, reduce wear and tear, etc. Is there any way to get the proper VCore power without disabling the voltage adjusments? Or is this something I shouldn't even care about if my system is 100% stable the way the AUTO setting volts it anyway?

 

[unclewebb]

Sounds like you should set it on AUTO and be happy. What program are you using to report core voltage?

 

[jdh79]

Speedfan, Ausus Probe, CPU-Z all report the same VCore numbers. I guess it's also possible the voltage reading on the board is misreporting the VCore on P5B's; it could be that is the issue rather than it actually undervolting.

 

[unclewebb]

Some of the E6600 / E6700 processors are designed to run at approximately 1.25 volts under load.

The E6300 / E6400 typically have a VID of 1.325 volts.

I think CoreTemp correctly reports what VID is built into your processor. I trust SpeedFan for correctly reporting core voltages.

If your computer runs 100% stable at full load, which it should, then let SpeedStep do its thing and save you a couple of watts.

 

[jdh79]

CoreTemp reports a VID of 1.325 under load. Is the VID the current VCore voltage, or just the spec voltage?

 

[unclewebb]

CoreTemp reports a VID of 1.325 under load. Is the VID the current VCore voltage, or just the spec voltage?

VID is just the spec voltage of the chip and not what it is currently running at.

I don't know if CoreTemp reports this value correctly or not.

 

[jdh79]

So, is there any way of telling if the VCORE that is being read in Windows is accurate and not just a reporting error without buying a multimeter type thing? Because if you boost it up about 7-8 one hundreds to account for the droop and the readings in Windows aren't correct, you potentially are overvolting.

Also, I ran SP2004 on both cores with the most CPU intensive settings for 8 hours last night without an error, and I have yet to have any problems whatsoever with stability (only exception was when I first installed Windows and it wouldn't install because the VDIMM was at 1.8 and the OCZ Platinum RAM I have requires 2.0) Do the Conroes just have enough fault tolerance to handle lower voltages, and it's something I shouldn't worry about?

My CPU temperature idles at around 34 CPU/41 both cores with a 32 case temperature
Under a normal heavy load (running Oblivion with several open apps) it will go up to about 44/50
If I use TAT 100% load the temperature hits 55/65.

This is with the stock HSF. Am I right not to be too worried about this when I am not overclocking, since using the stock HSF and stock settings means I am fully warrantied, and I am still coming in 20 C below the point of any thermal throttling? Plus, the TAT torture test isn't really realistic to anything you'd be doing, so realistically, about 50 C is the max I will hit.

I just don't really want to have to replace the HSF, possibly remove the board, and spend an additional $50 if at all avoidable.

 

[unclewebb]

I've been running the Intel heatsink and fan since day 1 without a problem while overclocking to 3400 MHz and sometimes beyond! If I ever decide to do some serious overclocking then I'll upgrade to a +$50 cooler.

If you are running stable then stop worrying. I've run my C2D from 1.1 volts to 1.55 volts and from 1000 MHz to 3700 MHz and pretty much every point in between these two extremes without any problems. These are great processors and extremely flexible so enjoy.