Shunt modding 3080 ti. Thoughts.

Andrew_Carr said:
just buy some conformal coating. I wouldn't use nail polish on a GPU due to the temps, and there's not much price difference anyway.
Click to expand...
Either would work just fine... the shunt resistors are far enough away that you would not see any problems with nail polish... I use it all the time when I dont have conformal coating or kapton tape.
 
Samuraijack23 said:
Better than ti at those settings
Click to expand...

By gains I meant total power limit increase not clocks........ Stock resistance + Decrease in resistance (via lm) = Final resistance reading from this we/you can calculate total GPU power limit. With the temps that your screenshots show you would be able to possibly get higher gains if the gpu was cooled a little bit better.

The correct way to tune for this mod is to watch precision x or afterburner and see if the gpu is bumping the power limit @100% gpu power if so increase the power limit in the oc software by 1%-2% until the card is no longer hitting the power limit. (The thing that you have to keep in mind is if you shunt mod the gpu with a 8mo resistor you are effectively increasing the power limit by 66% so 100% + 66% = 166%, This is out of the gate without any software modification) Anything beyond this EG. 115% in the settings is just creating heat and netting no gains.

This is why I asked you for a final resistance with your mod. There is alot more to it than just changing the resistance and increasing your power limit in afterburner/precision.

Either way it looks like it netted some nice gains.
 
I have ventus. Cooler sux dick. Work needs to be done.

I am volt limited more than anything at this stage in games I play.

Dat 4080 bich is no king. 💪
 
Doesn't seem that impressive when aftermarket 3080 Ti like the AORUS Xtreme can hit close to 2 GHz out of the box on a 400W power limit while running under 70°C on standard paste. 600W at 1965 MHz and 94°C looks like a poor overclocking attempt.
 
Armenius said:
Doesn't seem that impressive when aftermarket 3080 Ti like the AORUS Xtreme can hit close to 2 GHz out of the box on a 400W power limit while running under 70°C on standard paste. 600W at 1965 MHz and 94°C looks like a poor overclocking attempt.
Click to expand...
Not to mention it will burn up in a couple months.
 
Last edited:
Samuraijack23 said:
Under water....

None on air.
Plus your clocks at 4k in this game wont even reach 2k.
Click to expand...
index.png


Source: https://www.guru3d.com/articles_pages/evga_geforce_rtx_3080_ti_ftw3_ultra_review,32.html

This is an out of the box 3080ti FTW3 air cooled version.....

Another Example: Courtesy of thefpsreview.com
https://www.thefpsreview.com/2022/04/04/evga-geforce-rtx-3080-ti-ftw3-ultra-refview/9/

"The EVGA GeForce RTX 3080 Ti FTW3 ULTRA GAMING entered the match with a factory overclock that gave it a running start compared to the GeForce RTX 3080 Founders Edition. We were able to take it to the next level and stabilized it at an average 2047MHz and 21.8GHz memory. That makes the observed core clock 180MHz and the memory 12.8GHz faster than the stock EVGA GeForce RTX 3080 Ti FTW3 ULTRA GAMING clocks."


Samuraijack23 said:
I don't do benches.
Click to expand...

Thats cool but the story is starting to fall apart..... Your temps are holding you back...
 
Enigma said:
View attachment 537815

Source: https://www.guru3d.com/articles_pages/evga_geforce_rtx_3080_ti_ftw3_ultra_review,32.html

This is an out of the box 3080ti FTW3 air cooled version.....

Another Example: Courtesy of thefpsreview.com
https://www.thefpsreview.com/2022/04/04/evga-geforce-rtx-3080-ti-ftw3-ultra-refview/9/

"The EVGA GeForce RTX 3080 Ti FTW3 ULTRA GAMING entered the match with a factory overclock that gave it a running start compared to the GeForce RTX 3080 Founders Edition. We were able to take it to the next level and stabilized it at an average 2047MHz and 21.8GHz memory. That makes the observed core clock 180MHz and the memory 12.8GHz faster than the stock EVGA GeForce RTX 3080 Ti FTW3 ULTRA GAMING clocks."




Thats cool but the story is starting to fall apart..... Your temps are holding you back...
Click to expand...
The best part? It's running cooler than stock with that overclock and no liquid metal.

1672336668755.png
 
How did you get the correction factor for HWiNFO to display the actual power consumption after the LM mod? That's a tricky resistance to measure and seems difficult to correctly and precisely estimate
 
NattyKathy said:
How did you get the correction factor for HWiNFO to display the actual power consumption after the LM mod? That's a tricky resistance to measure and seems difficult to correctly and precisely estimate
Click to expand...
It wont show it you have to calculate it manually at least on a 2080ti, not sure about the 30 series...

Samuraijack23 said:
I didn't. It did itself. Like native
Click to expand...
I would be interested in a clamp ammeter reading on the 8pin connectors compared to the software reading side by side that would eliminate the software just guessing as to what it is putting out. That would be an easy calculation taking the reading and adding 75 watts to the number would give you total card wattage.
 
Enigma said:
It wont show it you have to calculate it manually at least on a 2080ti, not sure about the 30 series...
Click to expand...
This 100%. My question was a test for OP, I had to manually apply an offset in HWiNFO for my shuntmodded 3070Ti and 3090Ti as has everyone else who has done any variation of the mod. I don't think OP's mod is working right given the enormous power consumption being reported at low voltage and clocks but I don't think they're going to admit that. Unless the GA102 chips going into 3080Ti are incredibly low-bin compared to the ones going into 3090Ti or the high temps are causing some sort of thermal runaway triggered by leakage current, I cannot comprehend why this 3080Ti would be hitting 450-600W at less than 1V/2Ghz when my rather mid 3090Ti runs at 1100mV/2145Mhz all day at 350-500W and has never hit the modded 560W PL in-game.

Enigma said:
I would be interested in a clamp ammeter reading on the 8pin connectors compared to the software reading side by side that would eliminate the software just guessing as to what it is putting out. That would be an easy calculation taking the reading and adding 75 watts to the number would give you total card wattage.
Click to expand...
I would also be very interested to see this data.
 
Like ti. Three is a vid.

People need to realize that water isn't air. Plus at this point its a native power just like galax.

Need to see same game with settings. I provided all the info. Now it's up to u to test yourself and post it here.

Compare results.
 
Samuraijack23 said:
Like ti. Three is a vid.

People need to realize that water isn't air. Plus at this point its a native power just like galax.

Need to see same game with settings. I provided all the info. Now it's up to u to test yourself and post it here.

Compare results.
Click to expand...

I still call BS. I wont believe the power draw readings until I see a reading with a clamp meter and know what your final shunt resistance is on the gpu.

And here is why my 2080ti shunt modded is pulling north of 775watts with 2 8 pin connectors running 1440p in kombustor all at under 60C (as stated above you can put whatever offset you want in that osd):

1672671346285.png
 
I am downloading 3d mark now.
Will see how much I get off of it.

I upload a new vid running cod with 2k+ all the time
 
So, if you use LM there’s a good chance the card won’t go into 3d mode. Around the 2xxx series nVidia bounded the resistor for expected limits. There was a relatively tight band the power regulator expected.

I also don’t think you’d get real power readings. It’s the point of a shunt mod.
 
Last edited:
Dayaks said:
So, if you use LM there’s a good chance the card won’t go into 3d mode. Around the 2xxx series nVidia bounded the resistor for expected limits. There was a relatively tight band the power regulator expected.

I also don’t think you’d get real power readings. It’s the point of a shunt mod.
Click to expand...

Correct there is a limit to the shunt mod if you go much over a 8MO resistor soldered to the stack the card will lock to 2d clocks as a kind of safe mode and not clock above 300Mhz.

You are also correct that the card cannot sense the modded resistance of the shunt mod. We know through simple math and using a kill-a-watt meter what kind of base increase that we can get when using a 8MO resistor in the stack and that is roughly a 66% increase in the powerlimit when the card is stock with no software based mods.
 
You must log in or register to reply here.
Back
Top