Tualatin & ron VCore Question

[DSTA]

Hi all!

Simple question really for the Tualatin crowd:

does your VCore change depending on processor load, i.e. does it drop a bit when running full load? If so, by how much?

Reason I'm asking is that I'm trying to figure out how Intel's new VRM 8.5 spec works.

[Yodums]

It depends on your PSU.

It sometimes rises...

But if it were to drop it won't drop awhole lot maybe 0.1-0.3

Yodums

[DSTA]

Well, to better explain why I'm interested:

There's a diagram in the Tualaron spec sheet (page 30) that describes how VCC should be, depending on the current the CPU draws.

If I'm reading this right, it means that with higher CPU load (=more current drawn), the mobo's VRM has to deliver *less* VCC to the CPU to stay in Intel's spec.

Example figures from the spec go like this:

With a current of 10A, VCC has to be between -0.035V and +0.015V relative to the 1.475 "typical".

With a current of 30A, VCC has to be *lowered* and be in the range -0.135V to -0.085V.

Does not sound all that much, but since it's kicking in at load (for example when a stability test is run) it could affect overclocking.

Took a look at a VRM spec sheet (FAN5056MV85 from Fairchild Semi), and apparently the regulator chips (or at least the Fairchild part) have load sensing capability.

What might be interesting for the overclocker is that (at least for the FAN5056) the VCC "drop target" is adjustable via a resistor.

Hoping this makes any sense I'll ask again if this drop in VCC under load is (more?) noticable on Tualatin (i.e. VRM 8.5) mobos.

[6502kid]

I have had my P3-s in 2 different motherboards.

At the default core setting, the bios health screen and
Sandra motherboard info and other utilities all show
this thing at 1.424v give or take a couple of thousands...

MBM5.1 shows it at 1.39

Nothing I do seems to really change it except raising the core
in bios.

Been running Prime95 and surfing for a few hours now,
no change in voltage. cpu temp has gone up to 29c.
I can get it to 30-32 if I turn up the heat in the house, or
bump the corev a bit, but loads dont seem to bother this
chip none.

:p

[DSTA]

Originally posted by 6502kid
I have had my P3-s in 2 different motherboards.

At the default core setting, the bios health screen and
Sandra motherboard info and other utilities all show
this thing at 1.424v give or take a couple of thousands...

MBM5.1 shows it at 1.39

Nothing I do seems to really change it except raising the core
in bios.
:p

Well according to the spec sheets (VRM 8.5 design guidelines) the P3-S has got a different "power curve" than the 256K cache Tuallys, much flatter. Vcc drop is only -0.06V max.

I wonder how the mobo makers designed the regulator to adapt to the different requirements of P3-S and P3/Tualaron.




Edit:

You say that the temp does not change much WRT to processor load. Maybe the VRM design helps that, since it's dropping the Vcore a bit when load is high?

[outhouse]

My CPU works like this, and a CV drop under load is normal.

CV default 2.10

under load 2.06

the T may be different let us know how yours does.

[6502kid]

I think you are right about the P3-s acting a bit different
that the other chips.

I have popped my 800eb into both of these motherboards, and
got a much wider range on both CPU temp and core voltage.
(Nothing extreme. About what would be usual...)

I think intel puts that S designation on there for a reason.
Server. Stable. I had my HS fan STOP one time, and
this thing kept on running UT for at least an hour or so.
When I quit playing, I check my hardware monitor, and the
temps were only about 3c over what I usually got at load.

CPU fan rpms were at zero, so I powered down, and put a
new fan on. That was 2 months ago. Didnt seem to hurt
it any. Running 148fsb (1405) right now, and voltage stays
exactly where it does if I have it at 133. So do the temps.




:cool: :cool: :cool:

[DSTA]

Originally posted by outhouse
My CPU works like this, and a CV drop under load is normal.

CV default 2.10

under load 2.06

the T may be different let us know how yours does.

Yup, a bit of drop is normal, and, after reading a bit more about the subject, I think that's why they introduced that variable VCore with the new VRM spec.

When using a lot of amps at high processor load and then suddenly dropping to idle, a conventional voltage regulator tends to overshoot more. Especially with a poor PSU that might be a problem - or not, and it's Intel over engineering their stuff ;).

BTW I'll certainly let you know how my Tully does - once I can get one :mad: . They have shown up at a few webstores here, price is ok with roughly 120 Euros for a boxed 1.0, but it's not in stock yet.

[DSTA]

Originally posted by 6502kid


CPU fan rpms were at zero, so I powered down, and put a new fan on. That was 2 months ago. Didnt seem to hurt
it any. Running 148fsb (1405) right now, and voltage stays
exactly where it does if I have it at 133. So do the temps.

:cool: :cool: :cool:

Nice chip for sure! Wish I could afford one.

Since I don't have a life ATM ;) I dug a little deeper in Intel's spec sheets - I think the less pronounced "swing" with all Tullys has more to do with the higher current draw at idle (more leakage on the 0.13 process?) than with the dynamic Vcore curtesy of the new VRM specs.

NEways, the topic will probably get a bit more coverage since according to the OCers front page, the spec is similar to the P4 now.

Comparing the VCore on different FSBs is a good idea (should have thought of that) since current draw should be higher with higher FSB.

But (trying not to sound too paranoid here) I'm not so sure if the on board monitors can really be trusted. Perhaps the mobo makers thought it would look unsuitable to have VCore move around by ~ 0.150 (256K Tuallys) depending on load.

Ah well. Once I can decide what mobo to get for my 1.0A (and the blasted CPU makes it here) I'll try to do some measurements on the socket.