1070 GPU. DVI port removed, wire in its place, why? Stuck on low clocks too.

[GlenBrim]

I got this recently as-is. Gigabyte 1070 video card.

DVI port is removed. I don't know why.
A wire is soldered across two spots where it looks like the DVI port was. I don't know why.

I've read of people simply ripping out DVI ports and the rest of the cards still working fine. Never read anything about soldering a wire in there for some reason.

The card seems to work fine, except it's stuck in low clock speeds. While starting a game or benchmarking, the clocks may momentarily increase to around where they should be, then goes way back down and stays there. Voltage barely budges from low. 0.75v or whatever.

I've tried it in two different computers, 500w and 600w PSUs, yes, the 8-pin power cable was plugged into it. Installed fresh drivers a couple times. I've had an Asus 1070 in another computer for years that works great so am plenty familiar with them.

DVI port is removed.
The HDMI port apparently doesn't work at all.
All three displayports work.

I've tried forcing higher clocks with MSI Afterburner, and setting Max Performance in Nvidia driver utility, but they still do the quick spike up a bit and then stay back down. The main problem with this card is the low clock speeds, but the soldered wire is interesting too so figured I'd see if anyone had some ideas why the card is this way.

[McWhitePink]

When there is low load, why do you expect to have high clock speed?

[GlenBrim]

I don't.
"While starting a game or benchmarking, the clocks may momentarily increase to around where they should be, then goes way back down and stays there."

[McWhitePink]

are you able to monitor the temperature while loading it? Might be throttled by temperature.

[frysu36]

Check GPU-Z screen when running some benchmark in low res. especially TPD and power consumption for each 12V source like also check voltage drop in each of them during benchmarking. It could be a power monitoring ic aka INA3221 or similar one or even those 20ohm sensing resistors.

attached pic. from some random 1070Ti just for reference

[GlenBrim]

The temps are always low. And the card barely loads during the quick clock spike since the spike is so fast. It's practically non-existent, but I can see it happen with HWinfo logging it. Last I looked, max clock speed from the spike only goes to 1000 maybe 1200mhz. And then after the spike it continues running the benchmark or game way down at 350mhz or whatever.

I looked at the bios specs with GPU-Z, and the max clocks in it look normal. Figured look and see if the previous owner modded the bios to stay at low clocks for some reason. So that looks fine.

I'll mess with it more this weekend checking those voltages as best I can and really looking it over for bad solder connections and odd things. The main curiosity on it is still why is that wire soldered on there where the DVI port was. I've seen people add wires to GPUs so they can increase the voltage, but not where the DVI port was. And who takes a DVI port off anyway unless trying to turn a 2-slot card into a 1-slot card? Strange!

[прямо]

Looks like the wire is just connecting one ground point to another, from what seems to be the solder pads for the DVI port shielding/ground/support legs.

You can safely remove it.

[GlenBrim]

Unplugged the fans since some of the wires looked iffy, and the card sits around 40c anyway at these speeds.

12v rail power isn't great at 11.9x volts, but I think the minimum standard is way down around 11.4v.

Removed that ground wire. The two solder blobs each had two little solder "pegs" standing up, and each side of that wire was wrapped around them. So a mild tug and the wire came right off.

No change after any of that.

Watched the clock speeds and volts more. With Heaven bench open but not benching, the clocks often sit at 8xx, memory at 2000, power consumption goes to 27 watts, GPU Core Load is 2%.

Running Heaven bench, clocks sit at 139/405. Power consumption 16 watts. GPU Core Load 98%.

So, it shows more power consumption with Heaven open to the main settings page than it does when actually benching.

GPU voltage maxes about 0.775v.

I could control clock speeds and volts a bit with MSI Afterburner. Figured if it could do 1500 or so MHz at 0.800v or so, (like my other 1070 can do), the card would be usable. Nope. Tried 8xx MHz at 0.715v or so. One of the pictures below shows that setting, while benching, 0.675v, 139/405 clocks, 16 watts.

Thanks, everybody. Next time on it I'll check some ohm readings, and that will probably be it.

Oh, and if I'm just on the desktop and shake a window or similar, the GPU will spike to 1506 MHz momentarily. So it can go that high. Sits at 139 MHz while benching, though.

[McWhitePink]

Please try to summarize your TLDR.
I have difficulties to read through all the words.
Did you measure the resistance as frysue36 suggested?

[прямо]

If 12V at 11.9V is "bad" for you then honestly, I don't know what to say.

You live in the ideal perfect world which is completely different from the rest of us

[frysu36]

Do not look on voltage as probably 12V from all sources are ok. Check all power consumptions in GPU-Z this Hwinfo is nice but not reliable 🤷🏼♂️ for example if you have INA IC involved for power monitoring and is reporting incorrectly then this could happen. I don't like the idea of 0.000W at any given time even momentary. If you have oscilloscope then you could also check all phases PWM's and if PMIC is controlled via PWM VID then you can check it also I mean confirm if you have appropriate vcore on corresponding PWM VID level