Core voltage too low?

[sami_la]

My 2.4c is set @ 1.55v in the bios, and is running @3.4. CPUZ however reads the voltage @ 1.500, dipping as low as 1.450 and as high as 1.520
the bios is reading the volatge @ around 1.488, with fluctuations
sandra reads the voltage between 1.47 and 1.52

when i up the voltage in the bios (1.575v) in an attempt to clock higher while retaining stability, (3.5ghz), the bios reads an average of a measly 1.525volts

I have an enermax 350w supply (EG-865P), and its feeding my 9800, as well as 6 fans (one 12cm, two 8, one 7, one 6, one 4) Is it overworked?

I have an msi 865pe rev. A2 with latest bios version, could it be the problem?
My temps are fine, 28c as per sandra and 30c as per msi (idle and water cooled).

System posts @ 3.73(didnt try any higher as its pointless), however if i try to raise the voltage over the stock suggested 1.525 volts on the bios, (which is more like 1.45volts effective), i cant post past 3.5. Again my temps are fine so this whole behiavior is a mystery to me. More importantly, the fluctuations and starvation of my core voltage has been worrying me for almost a year now.

Ram is way under spec @ 378mhz (PC3700), as it would need to run @ 454 on a 5:4 multiplier, and i do plan on pushing the cpu further.

Any suggestions? And sorry for the looooooong post.

 

[Vio1]

What motherboard are you running off of? Sounds like you need a droop mod.

 

[sami_la]

i have an msi neo 2 865pe revision A2, bios version 2.5
I have also just discovered that when i force my cpu into semi-load, (by moving the cpu-z window violently accross the screen) my voltage invariably drops from 1.52 to 1.450 or at least 1.47

And what is a droop mod?

 

[sami_la]

also, @ 1.525 volts on the bios, i cant go into windows @ 3.5.
@ 1.55volts, i was cruzin for like 2hrs, did some sandra benchies then out of nowhere when the cpu was idle on top of it, the thing just reboots.
@ 1.5625 volts, it will post then choke.
@ 1.5375, its ok so far, but none of these values are real voltage values
1.525 is really 1.45-1.47 on idle, 1.5375 is 1.47-1.49 on idle, 1.55 is 1.49-1.52 on idle and 1.5625 is 1.5-1.53. But why would the system crash when the voltage is up to spec and when the temperature never exceeds 40c load? Undervolting is more stable?!!? neways, beats me.

 

[MysticX23]

Use another program to see if you have the same results. Speedfan shows Vcore as well.

 

[Lancelot]

It's your PSU trust me. I ran into problems with a Fortron 350W PSU not being able to maintain a 3.5Ghz OC with a 2.8C after I added an 9800pro...

 

[sami_la]

hmm, okay, i'll try the power supply thing, but before that, i remember reading that someone said somewhere in ocforums that you cant have 2 different power supplies running your motherboard and video card because the differential between the voltage between the agp slot and the actual vdo card might cause both supplies to fight against each other and eventually lead to death of vdo card. Is it ok if i just power them like that for a couple of testing hours?

 

[sami_la]

if that doesnt work, im willing to try the droop mod, if u guys help me out cause i dont know where to get plans or procedure for the msi neo2 865pe and also because it would be my first volt-mod.

 

[sami_la]

one more thing, my enermax 350w is rated @ 26A on the 12v rail, i thought it was enough. I will try nonetheless to use 2 pwr supplies and update u guys on the situation

 

[Falkentyne]

As far as I know, *all* i865 and i875p boards _will_ reduce the core voltage at load. It seems to be based on the vr-down specification at Intel.
And I BELIEVE the cpu's reduce the voltage internally, even more.

A droop mod is actually *violating* all Intel specifications.

However, the voltage reported by the BIOS may not really be the voltage being sent to the chip. What I recommend, is (besides getting higher than a 350W PSU, that is highly recommended), buying a decent digital multimeter and testing the actaul core voltage line, to see what voltage you are really getting at idle and at load. The BIOS may not be accurate.

Do that, and get the PSU.
Once you have a measurement of the real core voltage being sent to the processor, then we can take things from there.

One thing worthy of note:
the max VID for a 2.4C (so where you are not running voltage out of spec) is 1.525v. VID=voltage identification signal (the signal the mobo gives the CPU, i think...or is it the cpu giving the motherboard....)

Also, assuming you are using 1.525v in the BIOS, the lowest voltage the CPU will handle (it drops the voltage internally) is 1.345v, and highest is 1.425v. These voltage values have a tolerance of -5% and +10%. Take 1.425v, add 10% of that, and you get around the VID (1.525v). The VID is the max voltage that processor will ever use.

All the processors have a VID range, and if the voltage is kept within that vid range, they are in specification and will operate within a certain value (below the VID, as above). But I don't know how this works if the VID is exceeded.
There might be a formula, regardless of the VID, for how much of the VID the processors will actaully use.

Anyway, this is taken directly from the Intel spec docs for the P4C and EE.
As far as the effect of motherboards drooping the voltage, I'm not sure. Perhaps both (CPU droop (as above) and motherboard droop) come into play. But please check with a multimeter first.

You can either ask on the forums for how to get the core voltage measurement (someone will help you) or ask on xtremesystems.org.

Be sure to try a better PSU, though ; that IS important.

 

[sami_la]

ok, so i combined 2 psu to share the load, one (26A @ 12v) was strictly for the motherboard and vdo card, the other one dealt with the rest. Same fluctuations, same low voltage.

Before i go out and buy a digital multimetre, i have an analog one and has a 1.5v battery setting. I was wondering (but doubting) if it'll do the job?

thx for ur help