The fault is not in your processor, but in your motherboard

[AlHill]

Are you one of those who are constantly complaining about 13th and 14th gen stability and BSODs?  Well, read this:

https://videocardz.com/newz/asus-adds-intel-baseline-profile-to-its-z790-motherboards-amid-core-i9-stability-issues

While this specifically mentions ASUS, MSI and Gigabyte are in this mess also.

Doc (not an Intel employee or contractor)
[If you find any Intel driver you might need, download and save it now.]

[Keean]

> No....resistance of a metal increases not decreases .....the number is getting bigger....you are confusing something here....

Silicon is not a metal, it's a semiconductor. The junctions of the transistors are doped (either N or P doped) silicon.

[KrissyG]

look up the meaning of 'metal' and 'semiconductor' , you are again confusing terms. 

[Keean]

What are you asking me to look up? The resistance of silicon decreases with temperature. That's a fact.

[KrissyG]

---

@Keean wrote:
What are you asking me to look up? The resistance of silicon decreases with temperature. That's a fact.

---

yea....with negative temperature it decreases. 

"look up the meaning of 'metal' and 'semiconductor' " just do exactly that.

[Keean]

https://physics.bu.edu/~duffy/sc526_notes05/Rtemperature.html

I'll extract the relevant bits here:

> In some materials (like silicon) the temperature coefficient of resistivity is negative, meaning the resistance goes down as temperature increases. In such materials an increase in temperature can free more charge carriers, which would be associated with an increase in current.

[Keean]

Metal:
> In physics, a metal is generally regarded as any substance capable of conducting electricity at a temperature of absolute zero. Many elements and compounds that are not normally classified as metals become metallic under high pressures.

Semiconductor:
> materials which have a conductivity between conductors (generally metals) and nonconductors or insulators (such as most ceramics). Semiconductors can be pure elements, such as silicon or germanium, or compounds such as gallium arsenide or cadmium selenide.

[KrissyG]

---

@Keean wrote:
https://physics.bu.edu/~duffy/sc526_notes05/Rtemperature.html

I'll extract the relevant bits here:

> In some materials (like silicon) the temperature coefficient of resistivity is negative, meaning the resistance goes down as temperature increases. In such materials an increase in temperature can free more charge carriers, which would be associated with an increase in current.

---

Right? I am so dumb for cooling my CPU, it would have gotten much faster at 500°C. 
Read it again, does it say 'pure silicon' somewhere? or maybe it says extreme temperature? 

Now look for 'gate' in PMOS and NMOS and its chemical components. 

[Keean]

A field effect transistor is like a switch, the gate voltage controls the resistance between the source and drain. Current flows from source to drain through the substrate (silicon). No current flows from the gate at all.

So the resistance of the gate is irrelevant, what we actually care about is the capacitance of the gate.

The reason you cool your CPU is so you can run it faster.

In order for a 0 to change to a 1 the transistor has to charge the capacitance of its output. The higher the current the faster it can charge, so the higher clock speed is attainable.

Increasing the voltage increases the current, so let's you run faster, but you are limited by the heat dissipation of the silicon. The CPU itself throttles back the clock speed to protect itself when it gets too hot.

Once it gets too hot it will emit a puff of smoke and cease to function.

Early CPUs with no IHS and no self throttling would often fail like this more often less instantly if you forgot to attach the cooler.

[KrissyG]

at last you googled some.

Wish you googled some more. Gate resistance matters, infinite resistance would be a complete insulator....so an insulator on top of another insulator - taht would not work at all,
0 ohm would lead to destruction of the MOSFET. 
Temperature affects gate resistance....as well as the silicon

Another thing is the metals that connect the MOSFETs to the rails and other components.

in one of your previous comments you wrote "Power: is dependent on current" and now "Increasing the voltage increases the current" ....a sudden change of heart? lol

"The reason you cool your CPU is so you can run it faster." .....that literally say nothing about why a CPU needs to be kept cold for high frequency.
For a transistor, switching is work, the faster you work, the more you sweat.
It's as easy as that.

Reducing the voltage will allow higher friequency as the current is also lower....lower power....lower heat....lower resistance .
You mentioned capacitance of the gate, ironically, heating up a transostor , increases all components size.

Physical size is the enemy of frequency, teh smaller - the higher the freuquency,,,,also you would need less power to run it too.
There is more to it, as the size it becomeing extremely small, suddenly resistance and other factors actually increase....so there is a limit to high frequency at a too small size.

You can watch this video, the first half is what you need to hear.It's about MOSFET resistance https://www.youtube.com/watch?v=Apo8KTQJMB8

[Keean]

> Originally Answered: Why is the GATE current zero in FET? The DC current into a MOSFET's gate is 0, because the gate is basically insulate by silicon dioxide (glass). The gate forms a capacitor with the semiconductor channel between the source and drain

[Keean]

> You can watch this video, the first half is what you need to hear.It's about MOSFET resistance.

Yeah, that video says exactly what I was saying. Note he talks about resistance Rds that is the resistance between the drain and the source and it is through silicon, hence Rds will decrease with temperature.

> Reducing the voltage will allow higher friequency as the current is also lower....lower power....lower heat....lower resistance

No, reducing the voltage will result in a lower Ids (current between drain and source) and it will take longer to charge the output capacitance, and hence take longer to cross the 'on' threshold, and so will not be able to operate at such a high frequency.

Extreme cooling like liquid nitrogen works because there is so much power through the chip that the junction temperatures (Tj) are in the normal operating range for the chip, not at low temps, or it would actually be slower.

Basically what you need to be faster is more voltage, and you need to put the chip in an operating environment where that extra voltage does not cause it to fail instantly.

If you find that lowering the voltage makes your chip faster it's because it was thermal throttling (this is the chip lowering it's clock speed to keep the temperature under control due to too much current). If instead of lowering the voltage you improved the cooling (like with liquid nitrogen) it would actually go even faster at the higher voltage.

[KrissyG]

i wrote "so an insulator on top of another insulator " i wonder what would that be maybe one of those insulators is your SiO2? 

You wrote "No, reducing the voltage will result in a lower Ids (current between drain and source) and it will take longer to charge the output capacitance, and hence take longer to cross the 'on' threshold, and so will not be able to operate at such a high frequency."

"it will take longer to charge the output capacitance" with same current but lower voltatge, you can achieve quicer charging of your capacitance , that is how it works, also Google undervolting. 

You wrote: "Extreme cooling like liquid nitrogen works because there is so much power through the chip that the junction temperatures (Tj) are in the normal operating range for the chip, not at low temps, or it would actually be slower."

The 9GHz were achieved for about 2seconds or less, at -200°C with i think liquid hellium, but that does not matter with what. The temperature at the CPU was not even for a biref moment warmer than -150°C.

You wrote: "Basically what you need to be faster is more voltage, and you need to put the chip in an operating environment where that extra voltage does not cause it to fail instantly."

basically what you need is to prevent overheating, and force teh CPU to run at lower voltage that it would theoretically need to achieve 9GHz, and you can do taht bcoz the resistance was low....that is why they were able to do it at 1,85V and not 2V. altho at 2V something would break, even with super cooled CPU

[Keean]

> "so an insulator on top of another insulator " i wonder what would that be maybe one of those insulators is your SiO2?

The gate is metal, it sits on a thin layer of silicon dioxide which is an insulator, that sits on top of the substrate which is silicon (a semiconductor). The source and drain are areas of the substrate which are doped silicon, in the case of nmos N doped silicon. The dopant is an element from group 5 that has an extra electron, so when the the dopant is diffused into the silicon it results in extra electrons, hence n-type (electrons are negatively charged, the 'n' is for negative). The whole FET will sit in p-type silicon doped with an element from group three which will create extra 'holes' or positive charges.

Without any charge on the gate, there is a high resistance between the n-type source and drain and the p-type channel under the gate. It's like two diodes facing opposite directions. As diodes only permit current one way, back-to-back current can go neither way.

When voltage is applied to the gate, no current can flow to the source or drain because the gate is isolated by the silicon dioxide insulator layer, however the positive charge on the gate attracts negative charges to the under-side of the insulator creating a thin negative channel between the source and drain, permitting electrons to flow, and the higher the gate voltage, the wider the channel, and the more current flows between the source and drain.

So more voltage on the gate = more current between the source and drain.

If no current flows from the gate, why does it take current to switch on a FET? The answer is all the tiny metal wires form a capacitor. The longer the wires the higher the capacitance, and the more current it takes to charge that capacitor so the the gate voltage is high enough to switch on the FET. Also the closer together the wires the higher capacitance, which is why there are problems scaling up frequency on the newer smaller-scale features and this acts as a limit on Moore's law.

So it's a conductor on top of an insulator on top of a semiconductor. The source and drain are in the semiconductor.

As for why cooling allows some transistors (FETs) to switch faster, it's because the 'on-voltage' on the gate required to switch the FET on decreases with as it gets colder.

So you were right about cooler temps allowing faster speeds, but for the wrong reason. It's not because the resistance gets lower, it's because the FETs switch on at a lower voltage.

So colder temps increases the resistance of the FET in the 'off-state' reducing leakage current (which is a good thing) and increases current in the on-state.

Going back to my original point, get too hot and it will destroy the transistor, because as it gets hotter the resistance of the silicon decreases, so that even when the FET is off it conducts. At some temperature the decreasing resistance dominates the increasing switch-on voltage. I should not have called this 'thermal-runaway' though because that's used to refer to a specific phenomenon in bipolar junction transistors.

The reason a single transistor can get hot is because silicon is a poor conductor of heat compared to metals.

[KrissyG]

you did not get my sarcasm, but in the end you found out, there is a resistance.
Good for you i guess.

Now what is left for you, is to google the connection between capacity/resistance and AC voltage, and you will be home!

[Keean]

> you did not get my sarcasm, but in the end you found out, there is a resistance.

Well its complicated There are three effects happening, the resistance of the silicon decreases as temperature increase (increasing drain current with temperature), the average channel resistance of the FET increases as temperature increases (decreasing the drain current with temperature), and the p-n junction barrier potential decreases as temperature increases (increasing the drain current with temperature). It is apparently possible to design a FET with a zero temperature coefficient, so the drain current does not change with temperature at all in the operating temperature range. 

However I agree, you are right that the kind of FETs we are talking about here have an overall decrease in drain current as temperature increases.

> Now what is left for you, is to google the connection between capacity/resistance and AC voltage, and you will be home!

I hope I don't need to google that its been a while since I did my electronics degree, but most design these days is writing VHDL or Verilog code, and you just let the layout tools create the masks... This kind of stuff is more device physics.

I think you mean impedance not resistance?  The impedance of a capacitive circuit decreases with frequency, so higher frequencies require more current to charge the capacitance in less time. So to get more current for a higher frequency you need more voltage (to increase the drain current) or a lower temperature (to increase the drain current).

> Reducing the voltage will allow higher friequency as the current is also lower....lower power....lower heat....lower resistance

I don't think this is true, if the cooling remains constant, as the increase in voltage will result in a greater increase in drain current than the decrease in drain current due to the increase in temperature.  

To increase the frequency you can either reduce the temperature by cooling the chip better, or increase the voltage (until you destroy the chip, or thermal throttling circuity saves you).

If you increase the cooling a lot, you may be able to hit higher frequencies with less than stock voltage, but you would still be able to go even faster by increasing the voltage. Thats why the 9Ghz overclock needed 1.85V for this and not 1.2V, and they could not apply more voltage without it failing, or they would have gone even faster. 

Its easy to validate the chip runs slower with lower voltages, just disable turbo boost, and see how much it reduces the minimum voltage you can set and still have a stable CPU.

[KrissyG]

Maybe related to the topic here:

i have been bashing about the CPU, so now about the motherbopards, since i got a perfect comparison , bcoz the only thing that changes is only the motherboard.

B760 GAMING X DDR4
versus
Z790 AORUS ELITE DDR4

First of all, the B760 does not defomr the IHS of teh CPU that much, if at all.
The Z790 does so extremely deform teh CPU, i can even see the light shinging through if i lay down the water block on the CPU without thermal paste.

The reason i swapped the boards was VRM MOS temperature, here same load of 125W for at least 1 hour at same room temperature:

However, the idle power usage is not the same....B760 had only 3W idle TDP, while depending on BIOS version, the Z790 has 11 or 14W (at least that or higher)

I saw here users stating high idle power usage as well as temperatures, and now i know why.

Overall, Z790 while CPU on load, keeps the CPU temperatures lower, as the PCB of the motherboard does not heat up as much as on the B760.

[muhammadhammaf2424]

I do not see this as solution, this literally turns an i9 into an i7.

[KrissyG]

gen 12 is LGA 1700, so is gen 13 and 14....the actual chip extremely similar, what if there is no difference except from number of cores and cache? and all are actually gen 12, but since users buy new CPU more often, a new name has been given.

If you look at benchmarks and stress tests, an i9 is just a slightly overclocked i7. in case of 13900k and 14900ks, they let it OC beyond any resonable limits, just ot beat AMD. Unfortunately, AMD already won the CPU market, rebranding old CPUs as new ones only made the things worse for Intel.

[ahmad1999]

It sounds like you're referencing a computer hardware issue. If you're experiencing problems with your computer and suspect the motherboard might be the culprit, it could be related to various factors such as compatibility issues, physical damage, or malfunctions. Troubleshooting hardware problems often involves a process of elimination, starting with identifying the symptoms and then testing components individually to pinpoint the issue.

If you need assistance diagnosing or troubleshooting the problem further, feel free to provide more details, and I'll do my best to help!

[michealkors]

Thank you for the update. It sounds like the issue lies with the motherboard rather than the processor. I appreciate your insight and will look into it further.