![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
What is vdroop?
- Thread starter l3ender
- Start date
buffbiff21
Gawd
- Joined
- Sep 27, 2006
- Messages
- 1,014
You set VCore to 1.35. You run a heavy app and it goes down to 1.25 or 1.30. Depending on how much it goes down is how serious the droop is. Or it may fluctuate liek crazy.
Easy way to describe.
Remedy? Keep the mosfet area cooler or mod the board voiding warranty.
Easy way to describe.
Remedy? Keep the mosfet area cooler or mod the board voiding warranty.
BillParrish
Supreme [H]ardness
- Joined
- Aug 25, 2006
- Messages
- 7,519
OK quick and dirty, Everything is an engineering compromise. Nothing is perfect.
A perfect VRD (Voltage Regulator Down) circuit, which is the circuity on the board that takes the 12V from the power supply and creates the 1.325V or whatever needed for the CPU, is possible but would cost more than the board. These VRD circuits have to be able to keep the voltage steady under changes in load current that are tremendous. For example your poor VRD circuit is happily sitting there thinking happy thoughts as you web surf and the CPU is doing basically nothing and drawing an amp or two. Nothing much is changing and if it does its not happening fast or very much. All parts of the VRD especially the filtering capacitors have some resistance associated with them even if just a little, lets just pull crap from out of the air and say a Cap has .1 ohm of resistance. If the VRD is pushing 1 Amp through that Cap on the way to the CPU, the Cap has a voltage drop of .1 volt and the VRD has circuitry that will compensate for that "lost" .1V , no problem.
You decide to fire up F.E.A,R. and hit "play" or whatever, almost instantly the CPU utilization goes to close to 100% and the current draw of the CPU goes from 1 Amp to 50 Amps or more. In a 1/1000th of a heartbeat the voltage drop across that Cap goes from .1V to 5.0V ! (Obviously this is a somewhat overblown example but the basic theory is true. ) The regulation circuit will compensate but it cannot do it perfectly or instantly because all this is happening in micro seconds and everything is changing. Due to some very complex electronic design issues (back to the Nothing can be Perfect law) the regulation circuit will "overshoot" the desired voltage before it can stabilize. This causes short term over-voltage spikes that can harm the CPU. (The same thing happens in reverse in going from full load to little or no load on the CPU but at worst the "undershoot" just causes the CPU to lock up and you reboot wondering what the hell caused your machine to crash, no harm done)
Intel is aware of this problem and publishes specifications and guidelines for standard VRD design. One of the specifications intended to protect the CPU is for a certain amount of Vdroop to be designed in. Whatever voltage the CPU is requesting the VRD gives it a little less to provide some safety margin for "over-shoot" during big changes in load current. Board designers can use these recommended values or design their own VRD but still must meet the spec to be Intel certified. Thats why it varies between boards, its all about the VRD design.
So if you leave Vcore on "auto" the VRD will typically deliver slightly less voltage than the CPU specifies and most boards act the same way if you manually set the Vcore.
This difference between what is "dialed in" and what the CPU actually gets is what the designer decided to "subtract" for spike protection and is called Vdroop.
(those "lost" voltages across components in the VRD on the way to the CPU are called Vdrop and can also cause the voltage at the CPU to not match the bios setting but that is a result of crappy voltage regulation and not as a result of a designed in safety margin. )
BillParrish
Supreme [H]ardness
- Joined
- Aug 25, 2006
- Messages
- 7,519
+ 200 points for using search and things (designs) change over time. NP. It is an interesting subject that should be brought up every now and then.
Well we have problems there too. Your numbers seem pretty high 1.32V - 1.21 = 0.09V difference.
Couple of things. CPU-z is really reading a voltage sensor that reads the cpu voltage so some room for error there.
If you have any of the C1E and EIST stuff turned on it may be affecting the cpu voltage. Current Intel CPUs are designed to work over a range of voltages and the CPU will request the voltage it thinks it needs to do the job at hand through what is called the Vid. So the voltage fed the cpu can vary by design.
---------------------------------------------------
Lots of blah below but your question boils down, I think, to; Can I trust the CPU-z voltage reading? Yes and no. I absolutely cannot stand Gigabytes supplied Easy Tune utility for monitoring and OCing but in this case it is what you should use for voltage monitoring. Install it, check your voltages and see how they agree or not with cpu-z, play with voltages, take notes of the differences in readings with and without the bios options I mention below and then uninstall that resource pig crap. (Hmm, Bill, seems you really dont like that stuff do you, it is getting a little better you know. STFU its butt ugly, wth is that skin susposed to be anyway ? Space ship thingy ? What am I, 12 years old ? Real men OC in the bios anyway.)
-----------------------------------------------------
Blah blah I got tired of typing. Summary: All depends on board design and how well the manuf implemented voltage controls and if the bios actually does it correctly and what options are set.
I would think/hope that if you set the voltage in the bios to anything other than auto the board would disable this Vid circuitry between the CPU and the CPU voltage regulator, or lock the Vid to the voltage value you selected and thus lock the voltage to what you set (and be fairly close but errors still creep in) . But we have no way of knowing without doing some fairly heavy testing of voltages what is going on. Someone may have done this for your board I just do not know.
All I can suggest is you look at what cpu-z reports both at idle and under load and see how much it changes, if at all. Then go in and temporarly change it enough to make a difference but not too awful high say 1.375V and see what you get at load and idle. Do this with the C1E and EIST both on and off.
So without knowing exactly what was set how in the bios it is real hard to say. And investigation should be done on exactly what CPU-z is reading - The Vid bits for what the cpu is requesting or a sensor that is monitoring what the sensor thinks the cpu is actually receiving .
At stock settings (all the voltage changing options turned on) at full load the voltage can be as much as .154 volts off of the desired setting and just meet spec. (Table 5 on page 20)
http://download.intel.com/design/processor/datashts/318732.pdf
oh and my previous post above mentions Vdroop when Vdrop should have been used. The voltage loss across the VRD components is Vdrop and the built in "safety margin" for over shoot is Vdroop. I was obviously drunk.
Well we have problems there too. Your numbers seem pretty high 1.32V - 1.21 = 0.09V difference.
Couple of things. CPU-z is really reading a voltage sensor that reads the cpu voltage so some room for error there.
If you have any of the C1E and EIST stuff turned on it may be affecting the cpu voltage. Current Intel CPUs are designed to work over a range of voltages and the CPU will request the voltage it thinks it needs to do the job at hand through what is called the Vid. So the voltage fed the cpu can vary by design.
---------------------------------------------------
Lots of blah below but your question boils down, I think, to; Can I trust the CPU-z voltage reading? Yes and no. I absolutely cannot stand Gigabytes supplied Easy Tune utility for monitoring and OCing but in this case it is what you should use for voltage monitoring. Install it, check your voltages and see how they agree or not with cpu-z, play with voltages, take notes of the differences in readings with and without the bios options I mention below and then uninstall that resource pig crap. (Hmm, Bill, seems you really dont like that stuff do you, it is getting a little better you know. STFU its butt ugly, wth is that skin susposed to be anyway ? Space ship thingy ? What am I, 12 years old ? Real men OC in the bios anyway.)
-----------------------------------------------------
Blah blah I got tired of typing. Summary: All depends on board design and how well the manuf implemented voltage controls and if the bios actually does it correctly and what options are set.
I would think/hope that if you set the voltage in the bios to anything other than auto the board would disable this Vid circuitry between the CPU and the CPU voltage regulator, or lock the Vid to the voltage value you selected and thus lock the voltage to what you set (and be fairly close but errors still creep in) . But we have no way of knowing without doing some fairly heavy testing of voltages what is going on. Someone may have done this for your board I just do not know.
All I can suggest is you look at what cpu-z reports both at idle and under load and see how much it changes, if at all. Then go in and temporarly change it enough to make a difference but not too awful high say 1.375V and see what you get at load and idle. Do this with the C1E and EIST both on and off.
So without knowing exactly what was set how in the bios it is real hard to say. And investigation should be done on exactly what CPU-z is reading - The Vid bits for what the cpu is requesting or a sensor that is monitoring what the sensor thinks the cpu is actually receiving .
At stock settings (all the voltage changing options turned on) at full load the voltage can be as much as .154 volts off of the desired setting and just meet spec. (Table 5 on page 20)
http://download.intel.com/design/processor/datashts/318732.pdf
oh and my previous post above mentions Vdroop when Vdrop should have been used. The voltage loss across the VRD components is Vdrop and the built in "safety margin" for over shoot is Vdroop. I was obviously drunk.
Last edited:
BillParrish
Supreme [H]ardness
- Joined
- Aug 25, 2006
- Messages
- 7,519
Thanks.
My old P965 board does about the same thing. I love Gigabyte bios but it appears Asus is the winner for implementing the "give this voltage and dont mess with it" race.
My old P965 board does about the same thing. I love Gigabyte bios but it appears Asus is the winner for implementing the "give this voltage and dont mess with it" race.