Nvidia GeForce RTX 2070 SUPER and RTX 2080 SUPER

Yeah i know. I was just stating how bad the Vrms on the clevo were.

Normally vrms are designed to operate at 40-50% of their average current handling capability for peak efficiency.

 

Now, regarding the extra power mod, the VRM's used in the MSI laptops can(and most likely are) report the current via IMON pin, and it reports back directly to the controller chip, no its proved that shunt mod on the main shunt resistors does work thanks to @seanwee, so either the VBIOS doesn't limit the power that is being reported by the VRMs, or might come into play at higher percentage increases in power.

Given that @Khenglish talks about a secondary mod, is it truly needed, or is it due to the implementation by Clevo that makes it necessary?

 

I just did the same thing as the desktop shunt mods, just on a laptop.

What a shunt mod does is drop the resistance across the current monitoring shunt (5 mohm) by either soldering another shunt in parallel, shorting it with a wire or solder or just outright replacing it with a shunt resistor of lower resistance.

This works because of the way nvidia designs their power delivery circuit. There is a voltmeter across the shunt resistor which measures the voltage drop over the shunt resistor.

Then using Ohms Law ( I = V/R ) they can calculate the amount of current passing through the shunt resistor.

Example : 50mv/5mohm = 10A

Then this value is multiplied by the voltage going to the VRMs which is 12V so (P = I*V)

10A * 12V = 120W

Now when we do the shunt mod, we drop the resistance of the shunt resistor, but nvidia's circuitry doesn't know that and still thinks that it's 5mOhms.

A lower resistance will result in less voltage drop across the resistor. So let's say we are still using the same amount of power 120w and our shunt resistor has dropped to 3 mohms.

The voltage drop across the shunt will be : 10A * 3 mohms = 30mv

And the circuitry will do the same thing again

30mv / 5mohm = 6A 6A*12V = 72w So while the gpu is actually drawing 120w, the gpu will think that its only drawing 72w.

That's how the shunt mod works.

Tl:dr
Nvidia doesn't take the readings from the mosfets, they use a separate power monitoring circuitry which can be tricked via shunt modding.

 

Thanks for the explanation.
I know how it works, I'm just curious because @Khenglish made reference to a secondary mod, and given that the VRM's can also monitor their own current usage, I'm curious to know if this second current monitoring is also used to limit overall power usage, or if its only taken in account by the VRM controller to keep things in check.

 

I modded the current reporting resistor and didn't get any improvement. Still power limited despite reported power draw being less than half of the limit. This is commonly seen by desktop users who perform shunt mods as well, where there is some benefit from shunt mods, but not to the extent that is expected.

There is also a register set for each individual phases's current. This can be monitored and summed for an additional form of power limit. I'll try modding that next, but I want to wait until my extra FETs arrive since this is a more involved mod that I'd like to not revert just due to overheating. Something is limiting power and I can't think of what else it may be.

 

It's down to needing to use 19V VRMs.

 

Well yeah but there are too few of them as well.

 

MSI GT76 Titan DT 10SGS Laptop Review: Late Game Exotic Gear
notebookcheck.net

When the new and shiny fails vs its predecessor then it's time to stick the finger into the air and say enough is enough @jc_denton @Mr. Fox @Prema @Ashtrix @Arrrrbol +++

 

LOL my graphics score is 1000-2000 points higher.

 

To be fair the Timespy and 4K games score was higher

 

Yeah for sure. Its cpu has more cores.

 

My graphics is 28029 on a 2080 Super (150w)...

 

Is anyone interested in an RTX 2080 Super from Clevo!
I have a MXM card to submit.

https://www.ebay-kleinanzeigen.de/s...gb-x170sm-p870xx-p75-77xx/1545326786-278-4188