Why does my Vcore go up big time when my system is on Load?

[EPoX Tech]

The A64 memory controller is more intelligent than all of us - it will look at the capability of your actual memory and decide a suitable overall clock speed to memory divider whether it be /11 or 12 or whatever - the HTT and max memory clocks are only a reference for the decision made at POST by the A64 itself. You make your BIOS decision and live with the A64 decision itself. Remarkable to say the least.

To Candymancan - yes voltage will increase along side temps at 1.7 and thus 1.8 under load - that's the spec not the board - for more RPM from cooler suggest upgrading the cooler itself as the board will only ever supply c. .38A of current along 12V all day long as called for my the ATX spec for Fan headers.

[jnansley]

Originally Posted by EPoX Tech

The A64 memory controller is more intelligent than all of us - it will look at the capability of your actual memory and decide a suitable overall clock speed to memory divider whether it be /11 or 12 or whatever - the HTT and max memory clocks are only a reference for the decision made at POST by the A64 itself. You make your BIOS decision and live with the A64 decision itself. Remarkable to say the least.

So we're saying that the A64 is going to use an integral divider for the memory speed, no matter what, and that there are no asynchronous memory accesses? At least, not in the 3:2, 5:4, etc sense?

Originally Posted by EPoX Tech

To Candymancan - yes voltage will increase along side temps at 1.7 and thus 1.8 under load - that's the spec not the board - for more RPM from cooler suggest upgrading the cooler itself as the board will only ever supply c. .38A of current along 12V all day long as called for my the ATX spec for Fan headers.

So what's your personal opinion of thermally controlled fans?

Having the motherboard control the fan speed is a nifty idea, but I think it's asking for trouble. And I've always felt that you should be able to control both the "level" and "gain" of the voltage/temperature relation - haven't seen that in a BIOS yet. Every implementation I've seen falls into the "gimmick" catgory. Which includes most of the heatsinks with the thermistor sticking out of the side, rather than embedded in the metal.

[CandymanCan]

Well im ganna order a DD TDX waterblock so i can hook my water cooling system back up that was on my 2500+ system, iv been watercooling for 3 years now and going to heatsinks sucks to me.

Would hooking the cpu fan up to the molex extensions make the fan spin faster? You know the Molex pluggs that have the adapters for fans. Cuss i have one right here. I'll give it try but i dunno how fast the fan will be spinning cuss it isnt ganna plugged into the motherboard.

So the voltage go up and up on load when the temps get higher and higher? Thats stupide because the that will only make the temps get even higher. I really need my waterblock cuss the temps on this thing get to 56c-60c depending on how far i overclock.

[Gizmo]

Not a temp thing, a load thing.

Typically, the CPU only needs maximum voltage to be stable when you are pushing it hard.

As an example, my CPU needs 2.1v to be stable at 2654 MHz when fully loaded (running Folding@home, for example). However, if all I'm doing is sitting idle at the desktop, it is stable at 1.9v. So the logical question becomes, why not only run the system at high voltages when it is under heavy load? This is exactly what AMD has done with CNQ (and Intel with their equivalent). You increase the voltage to make the system able to run stably under heavy loads, and you decrease the voltage under light loads to reduce the heat output. That way, you get the best of both worlds; a system that runs cool, but still has plenty of oomf when you need it.

[CandymanCan]

Yea ok but what if 1.6v is stable under load @ say 2500mhz, but the thing keeps raising the voltage for no reason.

To me thats a bunch of crap. Even whe ni bring the cpu down to 8x200 which is 1600mhz it still wants to raise the vcore from 1.55 to 1.65

[EPoX Tech]

As we have stated there is a reason for this - it is within the specification of the PWM and the processor.

[Aedan]

Originally Posted by CandymanCan

Yea ok but what if 1.6v is stable under load @ say 2500mhz, but the thing keeps raising the voltage for no reason.

It appears to be the way that AMD designed the CPU, in that the CPU itself changes the voltage ID pins in order to request a higher voltage when under load. When the CPU is idle, it seems to change the voltage ID pins to request a lower voltage. The documentation I've read hasn't explicitly said if this is a hardware function of the CPU or if the BIOS has to provide this functionality. Either way around, it's required functionality to support the AMD64 processors.

[Aedan]

Originally Posted by CandymanCan

Would hooking the cpu fan up to the molex extensions make the fan spin faster? You know the Molex pluggs that have the adapters for fans. Cuss i have one right here. I'll give it try but i dunno how fast the fan will be spinning cuss it isnt ganna plugged into the motherboard.

I'd doubt it, unless the motherboard is controlling the fan speed. If you want a high speed fan, you'll probably have to fit one yourself. As far as ET's comment, that seems to point at why AMD did not equip a 7000RPM fan on the stock heatsink - it might exceed the current draw permitted by the fan headers on a standard no-name generic board.