Question Three 8-pin connectors for GTX 3080

[holmesc]

So i've been using two 8-pin connectors for my old 1080 and am moving on to a 3080 pretty soon. I have an old 8 -pin connector that I never used and wanted to ask if there will be any indicators that it's functioning properly (or improperly) once plugged in.

I don't know if it'll still work under two 8 pin connectors and slowly get damaged so i'd like to ask if I can do a test or see some kind of indicator that all three 8 pin connectors are fully functional.

 

[Dark Lord of Tech]

What PSU make and model?

 

[JeffreyP55]

I use splitters for my RTX3080ti. However it is a 1200w supply with a 1175w 12v rail. Feeling pretty safe. You need to have a decent supply with at least 750w. Still umm, questionable.

 

[Dark Lord of Tech]

What make and model ?

 

[drea.drechsler]

So i've been using two 8-pin connectors for my old 1080 and am moving on to a 3080 pretty soon. I have an old 8 -pin connector that I never used and wanted to ask if there will be any indicators that it's functioning properly (or improperly) once plugged in.

I don't know if it'll still work under two 8 pin connectors and slowly get damaged so i'd like to ask if I can do a test or see some kind of indicator that all three 8 pin connectors are fully functional.

My guess is you'll need all 24 pins connected properly for the GPU to initialize.

Ideally, it should be three 8 pin connectors on 3 cables all the way back to the PSU. That means either three complete cables into the PSU or three connectors at the PSU if modular. What that means is the PSU also should be designed to handle the power draw of that beast. Cheaping out on doubled up splitters won't help a PSU designed to power one 8 pin connector cut the mustard.

And even if it does have the power handling a splitter leaves a single cable to carry the current load of both 8 pin connectors from the PSU to the splitter. If made with 16 ga wire it would work OK, if 18 ga maybe but many PSU's are made with 20 ga., some 22 ga. even. The lighter you go with wire gauge the more the cable will impose it's own voltage drop that varies under load. That will challenge the line-side regulation of the card's VRM's.

As far as testing your cable: you simply need a bug light to test continuity from the PSU connector to the GPU connector. You can compare to your other cables, which is a gold standard and probably the best way. Or use a diagram like the below one linked for corsair PSU cables.

https://mainframecustom.com/wp-content/uploads/2019/11/Example-Diagram-1024x871.jpg

 

[Karadjgne]

The lighter you go with wire gauge the more the cable will impose it's own voltage drop that varies under load.

Naa. That cable is far too short to affect voltage. It takes 25ft or more. What the guage does affect is resistance. Electricity travels on the outer edge of wire, not the inside. The smaller the guage, the smaller the surface area, the higher the resistance to amperage. That creates heat and more resistance in a viscous cycle raising it until the wire burns.

DarkLords question is the only thing that matters. What psu? That's important a) for wattage considerations b) connectors. An 'old 8pin' is meaningless and potentially useless, especially if it's an unused EPS 8pin and not a pcie 6+2pin or if the '8pin' is psu side connection.

 

[drea.drechsler]

Naa. That cable is far too short to affect voltage. It takes 25ft or more. What the guage does affect is resistance....
...

That's exactly what I'm describing. Ohms law...E=I*R...resistance to current flow results in voltage drop. Basic electronics.

PSU power handling is important, of course, not discounting that. But the often overlooked effect is the voltage drop between PSU and GPU due to cable resistance that can causes wild voltage fluctuations with GPU loading during a game. Even if not great in magnitude it's very rapid and the effect is multiplied by the way a buck regulator (the VRM on the board) has to work...it's called line side regulation. Anyone wanting assurance of stability will try to make sure they have not just adequate power but a stable voltage supply to the GPU so the VRM isn't being challenged.

In extreme cases (like one of those 22ga super cheap PSU cables) the cable wires can get hot (P=I^2R). If bundled up tight that heat just builds until something melts and then shorts out. Kind of a nightmare scenario, thankfully not something that happens too often based on the lack of any posts I've ever seen where it's happened.

Dont get confused by the skin effect of current flow: that only affects AC and this is DC flow.

 

[Karadjgne]

In extreme cases (like one of those 22ga super cheap PSU cables) the cable wires can get hot (P=I^2R). If bundled up tight that heat just builds until something melts and then shorts out. Kind of a nightmare scenario, thankfully not something that happens too often based on the lack of any posts I've ever seen where it's happened.

Usually happens when ppl use adapters, especially Sata to 6+2 pin or molex to 2x 6pin etc.

P=I²R is only part of it. The gpu demands a certain amount of power to function, so that's a constant. Both sides of the = must remain the same, so as resistance builds, amperage drops at the receiving end, so amperage draw is increased at the source to compensate for the loss. As the amperage goes up, heat goes up, which increases resistance, then amperage, rinse and repeat in an ever climbing cycle until the resistance and amperage is great enough to create enough heat to melt the weak point, and you get a fire in the pc.

VRM's and gpu are no different than VRM's and cpu, there's going to be voltage fluctuations because of loads, but that's all board related, any voltage concerns as supplied by the psu are no different either, voltage isn't affected by guage when talking about wires a foot long. Even 22ga supply wires. The longer the wire, and smaller the guage, the higher the resistance.

A 10ft length of 18ga has a 1.06% voltage drop at 12vDC, or basically 0.128v. Most psu supply are 12-18", so figure a voltage drop of @ 0.013v, 12v being 11.987v ± at the gpu. That's moot, especially considering there really aren't any 12.0v outputs as a constant value, most psus range from @ 11.8v to 12.1v, 12v really is more a designation than actual output.

 

[drea.drechsler]

Usually happens when ppl use adapters, especially Sata to 6+2 pin or molex to 2x 6pin etc.

Indeed..and that's what I'm talking about; the "Y" adapters commonly used to make a single 8 pin compatible with a dual 8 pin card (or a 3rd, as it were). But this also includes using the "pigtail" second 8 pin on many PSU to GPU cables as well.

You have to know quite well what your card's power draw is capable of when using one of those...and as well the gauge and current capacity of your cable. And know how to convert power draw to current draw, then divide out the total current draw among the total number of wires to determine if your overloading and getting into trouble.

Or just follow some simple rules: like three 8 pins, three cables all the way back to a quality PSU with sufficient power handling capacity.

 

[holmesc]

What PSU make and model?

I currently have a EVGA SuperNova 750W 80 PLUS Platinum.
The extra 8 pin spoke of comes from that box.

 

[drea.drechsler]

I currently have a EVGA SuperNova 750W 80 PLUS Platinum.
The extra 8 pin spoke of comes from that box.

I'll leave it to the PSU experts to pass judgement on an EVGA SuperNova, but I'm certain 750W is sufficient for a 3080.

If the cable came packaged with the PSU then this is it's intended purpose. Why the concern whether it will function properly?

 

[holmesc]

I'll leave it to the PSU experts to pass judgement on an EVGA SuperNova, but I'm certain 750W is sufficient for a 3080.

If the cable came packaged with the PSU then this is it's intended purpose. Why the concern whether it will function properly?

I've had cables that malfunctioned in the past, so i'm worried it'll harm the gpu if I don't pay attention.