koroem
So since Jacob didn't provide much info with his statement, is there at least a set group of free downloadable apps and settings we can agree upon that should produce the desired results to compare against? There has been a lot of stuff mentioned through 4700+ posts here. Can we narrow that down to one or two? Should we be using GPUz or HWiNFO? I've seen both report different numbers? Should we be using Heaven or Superposition or something else for max power draw that should be easily measurable? People have reported problems using Precision X1, should we agree to use Afterburner?
 
I know its a tall order to get people to agree on this stuff, but maybe we could make a template that people can copy and post with results from the testing, the response from EVGA, and the before/after results that occur from this new Email RMA system they are introducing. Lets get organized and standardized people.

Attached Image(s)

koroem

TheNaitsyrk

koroem
So since Jacob didn't provide much info with his statement, is there at least a set group of free downloadable apps and settings we can agree upon that should produce the desired results to compare against? There has been a lot of stuff mentioned through 4700+ posts here. Can we narrow that down to one or two? Should we be using GPUz or HWiNFO? I've seen both report different numbers? Should we be using Heaven or Superposition or something else for max power draw that should be easily measurable? People have reported problems using Precision X1, should we agree to use Afterburner?
 
I know its a tall order to get people to agree on this stuff, but maybe we could make a template that people can copy and post with results from the testing, the response from EVGA, and the before/after results that occur from this new Email RMA system they are introducing. Lets get organized and standardized people.

I then


Don't mind starting this. GPU-Z seems to provide more accurate results than any other medium. Heaven 4 should push GPU past 119% if OCed as that's how my 2080 Tis behaved. 2560x1440 at ultra settings inc. tessellation etc. cranked to full is a good benchmark. 
 
I got them at home will post once I can.
 
 

What I'm doing is running opening GPU-Z to the sensors page, opening Heaven with DX11, Ultra quality, extreme tessellation, stereo 3d disabled, multimonitor disabled, anti-aliasing 8x, no full screen, resolution system (2560x1440p). I run heaven for 5 mins with GPU-z overtop of it, and while it it still open, in GPU-z I switch the measurements for GPU load, PCIe slow power, and all 8pin power/voltages to display average. I then press the reset button on the gpu-z sensors window and continue  run Heaven another 5 mins. This is to clear out all the lower numbers from when GPU-z was open without anything running, and only shows numbers from Heaven open and running.
 
If anyone has a better method, let me know. I just reflashed back to to stock bios I had saved months ago before switching to this XOC one. Tested that, now I've flashed the XoC one back on the card and beginning testing. Initial testing confirms what I kinda already knew, there is almost no difference between them, but will update in a bit I guess.
 
EDIT, Does Heaven increase its load when running the actual benchmark? Or is it the same draw on its endless "idle" loop?

koroem
Quick morning testing results, details on how test was done in post above. Assuming I should be seeing better numbers with the XOC bios eh?
 

 
 

TheNaitsyrk
Is that after Heaven 4? Works fine except that draw from PCI-E.

TheNaitsyrk

arestavo
Cards power limited on the normal and XOC VBIOS will be power limited on the Kingpin 520W ln2 VBIOS. People using the 1000W unlocked warranty voiding Kingpin VBIOS are brave and or have money to burn, and shouldn't post those results if they want to keep their warranty.

I don't see a problem. Cards are broken people are trying everything to try to fix. That's not what they paid for. Also, people here would need their serial numbers anyway which they aren't providing so they are safe.

q437637
so email for RMA info? I was hoping for a bios update or something, plus I have the 3985 not 3987 .. damn
 
And I put waterblock on. I kept all the original thermal pads/foam but I am not certain if it can be put together as before due to those foams on the VRM. They probably changed size, height after being taken off. Anyone able to put it back after installing a waterblock?

changboy
They want you believe they do something, but actually they do worst by telling you : YOU CAN RMA JUST ONCE ! This mean they just protect them self for multiple rma.

Kylearan

hyuiop
Would it not almost be easier to 16mΩ shunt the PICe slot and 8mΩ shunt the three rails to remove the power limitations and balance the power draws back to the rails from the card slot?

This doesn't fix the power draw whatsoever.
All it would do is change what is reported to the BIOS, so that the card wouldn't hit the limit of an internal power rail (not to be confused with TDP limit).
The power draw would still be messed up.  In fact it would even more messed up (e.g. PCIE 81W would end up maybe 90-95W but might appear to the BIOS as 60W).  It just wouldn't be reported as messed up.
 
You also have to be careful about NOT shunting certain shunts.  It's best that all shunts are shunted, but this applies to cards with fully digital controllers.  e.g. on a 3090 FE, if you shunt all the shunts EXCEPT PCIE Slot, it will cause MVDDC's power draw (reported, not real) to reach over 200W, causing a rail throttle on both MVDDC and PCIE Slot.  No idea about the eVGA cards.