Is a 3x 8pin to 12pin GPU power cable feasible?

[Rent-A-Worker]

After seeing the 2x and 1x 8pin to 12pin cable for the 30 series gpu, I was wondering if its feasible to make a 3x 8pin to 12pin cable.

 

Would such a cable (plus the proper cooling and bios) allow to founder's edition draw more power like the 3x 8pin STRIX or FTW3 gpu?

 

Or would I have a fire in my pc instead...

[Abdullah Bhutta]

29 minutes ago, Rent-A-Worker said:

After seeing the 2x and 1x 8pin to 12pin cable for the 30 series gpu, I was wondering if its feasible to make a 3x 8pin to 12pin cable.

 

Would such a cable (plus the proper cooling and bios) allow to founder's edition draw more power like the 3x 8pin STRIX or FTW3 gpu?

 

Or would I have a fire in my pc instead...

You would have a fire.

If you want more 8 pins then go buy a STRIX or FTW3

[Rent-A-Worker]

1 hour ago, Abdullah Bhutta said:

You would have a fire.

If you want more 8 pins then go buy a STRIX or FTW3

True, it's better to get purpose build cards that can handle the higher power draw.

 

Although I am still not sure why Nvidia then used the same 12 pin for the 3090/3080 as in the 3070, as one could also then use a 2x 8pin to a 3070

 

I remember something about more parallel wires should reduce the overall resistance of the wire? (foggy memories of high school education). Not sure if that's a plus if you use the stock bios

[Alex Atkin UK]

20 minutes ago, Rent-A-Worker said:

True, it's better to get purpose build cards that can handle the higher power draw.

 

Although I am still not sure why Nvidia then used the same 12 pin for the 3090/3080 as in the 3070, as one could also then use a 2x 8pin to a 3070

 

I remember something about more parallel wires should reduce the overall resistance of the wire? (foggy memories of high school education). Not sure if that's a plus if you use the stock bios

Only half the pins are connected on the 3070 if you look at the adapter.

[Rent-A-Worker]

17 minutes ago, Alex Atkin UK said:

Only half the pins are connected on the 3070 if you look at the adapter.

Hmm I can't find a clear image of the adapter, but from GN teardown of the 3070 it looks like all the 12pin are connected to the board.

[GoldenLag]

1 hour ago, Rent-A-Worker said:

Although I am still not sure why Nvidia then used the same 12 pin for the 3090/3080 as in the 3070, as one could also then use a 2x 8pin to a 3070

There is no perticular reason for the 12-pin to exist. it doesnt offer special current capabilties. and Nvidia decided to spec choke it with adapters. 

 

one can use a single 6+2 pin on the 3070. and you could use a single 8-pin EPS for the 3090 and 3080 while offering more current capability than the 12-pin. while also having the advantage of standardizing 12v powercables. 

[Guest]

3 hours ago, GoldenLag said:

There is no perticular reason for the 12-pin to exist.

PCB footprint is reduced.

 

[GoldenLag]

26 minutes ago, jonnyGURU said:

PCB footprint is reduced.

 

something the 8-pin EPS would also a accomplish. while also standardizing the powerconnectors. while its a perk of the 8-pin. its not something that isnt allready accomplished. 

[Alex Atkin UK]

13 minutes ago, GoldenLag said:

something the 8-pin EPS would also a accomplish. while also standardizing the powerconnectors. while its a perk of the 8-pin. its not something that isnt allready accomplished. 

What's EPS got to do with anything?  PCI-e power is not the same as EPS.

The point is, the 3090/3080 connectors take up less space than two PCIe 8 pin connectors and can handle more current.  I'd argue it looks a lot neater, at least if the adapter was long enough to reach the other side of the case be hidden away.

They probably used it on the 3070 too because its cheaper to stick with a single connector than to swap to something else.

[Alex Atkin UK]

4 hours ago, Rent-A-Worker said:

Hmm I can't find a clear image of the adapter, but from GN teardown of the 3070 it looks like all the 12pin are connected to the board.

 

[Guest]

34 minutes ago, GoldenLag said:

something the 8-pin EPS would also a accomplish. while also standardizing the powerconnectors. while its a perk of the 8-pin. its not something that isnt allready accomplished. 

The EPS 8-pin is still a larger connector.

 

And the 12-pin is also capable of carrying more current.

 

Or are you talking about how the 3070 doesn't have all of the wires populated on the 12-pin?

 

Maybe they're using the same PCB?

 

 

[GoldenLag]

12 minutes ago, jonnyGURU said:

Or are you talking about how the 3070 doesn't have all of the wires populated on the 12-pin?

im reffering to gimping it by using adapters that dont exceed 300w of current capability per spec (i know 6+2 pin is underspeced connector for its wiregauge)

13 minutes ago, jonnyGURU said:

The EPS 8-pin is still a larger connector.

its a negligable difference for the advantage of it allready being in widespread use. 

13 minutes ago, jonnyGURU said:

And the 12-pin is also capable of carrying more current.

indeed it does on paper, but currently they dont seem to be taking advantage of that. currently a EPS connector could do that same job. 

 

i just really dont like them introdusing a new Connector when existing solutions can do the same job. 

 

32 minutes ago, Alex Atkin UK said:

What's EPS got to do with anything?  PCI-e power is not the same as EPS.

exactly, its not the same as the PCIe powercables, but its effectively the same size while being in widespread use and has way more current handling per spec. 

 

33 minutes ago, Alex Atkin UK said:

The point is, the 3090/3080 connectors take up less space than two PCIe 8 pin connectors and can handle more current.  I'd argue it looks a lot neater, at least if the adapter was long enough to reach the other side of the case be hidden away.

and you can do the same with a single 8-pin EPS. and it takes up essentially the same footprint. bonus, the 4-pin also exists for lower power solutions. 

35 minutes ago, Alex Atkin UK said:

They probably used it on the 3070 too because its cheaper to stick with a single connector than to swap to something else.

i mean if cost was the reason behind it. it would indeed be cheaper to stick with a single connector. a connector used on existing hardware. aka the EPS 8-pin

[Guest]

2 minutes ago, GoldenLag said:

im reffering to gimping it by using adapters that dont exceed 300w of current capability per spec (i know 6+2 pin is underspeced connector for its wiregauge)

It's not "underrated".

 

The thing people seem to forget is that the current rating of a pin or wire is not referring to the maximum amount of current the pin or wire can deliver without consequence.  It's the amount of current a pin or wire can deliver SAFELY.  In other words, exceed that rating and things start to melt.  As the current increases, so does the resistance.  Voltage WILL drop all the way up until that "current rating".  The resistance creates heat.  Which, in turn, creates more resistance.  This also leads to another phenomenon.  Too many conductors too close to each other need to have their current rating derated because of the heating from adjacent/surrounding pins.

 

Just have a look at any Molex data sheet.  You'll see something like this:

 

 

Note how the more conductors OF THE SAME KIND has a reduced maximum current rating.

 

So just because an EPS12V connector, on paper, theoretically can "support" upwards of 28A doesn't mean you want to be anywhere NEAR that.

[GoldenLag]

24 minutes ago, jonnyGURU said:

So just because an EPS12V connector, on paper, theoretically can "support" upwards of 28A doesn't mean you want to be anywhere NEAR that.

im working off the basis that a EPS 8-pin can do 32a total with then a total wattage of 384w. taken from AHOC when he reffers to the EPS 12v. more specifically 8a per conductor. 

 

edit: which looking at you sheet there is correct with margins, or wrong by 0.3a if im reading it wrong. in which i will happily correct that. 

24 minutes ago, jonnyGURU said:

It's not "underrated".

so the 150w that the 6+2 pin can deliver per spec is what one should expect for that connector?

 

and i am aware of the aspect of voltagedrop that occurs. 

[Alex Atkin UK]

17 minutes ago, GoldenLag said:

im working off the basis that a EPS 8-pin can do 32a total with then a total wattage of 384w. taken from AHOC when he reffers to the EPS 12v. more specifically 8a per conductor. 

 

so the 150w that the 6+2 pin can deliver per spec is what one should expect for that connector?

 

and i am aware of the aspect of voltagedrop that occurs. 

I still don't understand your obsession with the EPS connector.  Using EPS would be no more useful than the new custom connector, as neither is available as standard on a PSU for GPU use.

Plus the whole point is to not have a CPU power connector and PCIe device connector be the same fitting, so its easier for people to know which connector goes where.

IMO having a single new connector that works for ALL PSU wattages would be very much a good thing.  I'm really not a fan of the PCIe connectors, they never seem to plug in as firmly as I'd like, especially with them usually being 6+2.

[Guest]

10 minutes ago, GoldenLag said:

im working off the basis that a EPS 8-pin can do 32a total with then a total wattage of 384w. taken from AHOC when he reffers to the EPS 12v. more specifically 8a per conductor.

I'm working off 7A per conductor as per the paper spec.

 

And a link to where AHOC pulls 384W from an 8-pin EPS12V with actual voltage, current and thermal measurements would be appreciated, because when >>I<< do this, I see SIGNIFICANT voltage drop and very high temperatures.  32A ≠ 384W because the voltage drops significantly at those high loads.  So I'm calling B.S.

 

12 minutes ago, GoldenLag said:

so the 150w that the 6+2 pin can deliver per spec is what one should expect for that connector?

 

and i am aware of the aspect of voltagedrop that occurs. 

I thought you were talking about the paper spec for the connector.  

 

150W is defined because that is the threshold of when voltages start to take a nose dive.  On paper, the connector can support 24A (same pins as EPS12V), but again, you don't want to be anywhere near that.

[GoldenLag]

4 minutes ago, jonnyGURU said:

And a link to where AHOC pulls 384W from an 8-pin EPS12V with actual voltage, current and thermal measurements would be appreciated, because when >>I<< do this, I see SIGNIFICANT voltage drop and very high temperatures.

he uses it has a theoretical when looking at PCBs, dont know if he has  CPU overclocking video for this. And i assume he doesnt have footage of this. 

 

5 minutes ago, jonnyGURU said:

I'm working off 7A per conductor as per the paper spec.

isnt that sheet for dual row.?

 

and the EPS is 4 rows of 2. making is 9.3a (in the sheet). Going down to 8a. (ignoring solid pins, since those would be higher spec)

7 minutes ago, jonnyGURU said:

32A ≠ 384W because the voltage drops significantly at those high loads.  So I'm calling B.S.

yeah i can certainly see that being the case. tho i would be interested to see the voltage drop at that current draw. 

9 minutes ago, jonnyGURU said:

150W is defined because that is the threshold of when voltages start to take a nose dive.  On paper, the connector can support 24A (same pins as EPS12V), but again, you don't want to be anywhere near that.

didnt know that the voltage took a nosedive after that. nice to know. 

 

 

12 minutes ago, Alex Atkin UK said:

Using EPS would be no more useful than the new custom connector, as neither is available as standard on a PSU for GPU use.

well one is allready in use, the other one isnt. so there is that. 

 

i am all for swapping out the entirety of the current connectors, because honestly its kinda a mess right now

13 minutes ago, Alex Atkin UK said:

Plus the whole point is to not have a CPU power connector and PCIe device connector be the same fitting, so its easier for people to know which connector goes where.

well you wouldnt have to wonder which goes where if they are all the same. and arguably the 6+2 pin and 8-pin EPS are confusing as they look very similar appearance wise. 

 

May 12vo save us all. 

[Guest]

15 minutes ago, GoldenLag said:

isnt that sheet for dual row.?

Yes.  Also make sure you're looking at the spec for the correct pin material.  Nobody is using the gold plated connectors.

 

[GoldenLag]

17 minutes ago, jonnyGURU said:

Yes.  Also make sure you're looking at the spec for the correct pin material.  Nobody is using the gold plated connectors.

 

yeah im using the sheet you presesented which is copper wire with tin plating. 

[akio123008]

34 minutes ago, GoldenLag said:

May 12vo save us all. 

On that note, why have multi-pin connectors at all?

 

I mean, if the purpose is just to get loads of power to components, and there's only one voltage used (as is done with 12vo) then why not just use a two-pin connector for everything? (something similar to XT90 or EC5) That would also help quite a bit with cable management.

 

I like the idea 12vo, but I just don't see why they wouldn't go a bit further and just let go of the old atx-ish style.

 

[Mateyyy]

1 hour ago, jonnyGURU said:

So just because an EPS12V connector, on paper, theoretically can "support" upwards of 28A doesn't mean you want to be anywhere NEAR that.

Roughly up to what current draw would you say is adequate for an EPS12V, or in other words, what would you consider not to be "anywhere near" that 28A mark?

 

I'm asking because the RTX A6000 takes an 8-pin EPS for power, and that's a 300W card.

[GoldenLag]

7 minutes ago, Mateyyy said:

Roughly up to what current draw would you say is adequate for an EPS12V, or in other words, what would you consider not to be "anywhere near" that 28A mark?

i mean the sheet he provided says at least 32a if i read it correctly. 

 

8a per pair. and there being 4 pairs. (edit: it says 9.3a but im just using my 8a as low ball)

[Guest]

1 hour ago, GoldenLag said:

yeah im using the sheet you presesented which is copper wire with tin plating. 

Oh.. That was snipped from a Micro-Fit+ data sheet.  I happened to have one open while reading this thread, so just used it as a point of reference.

48 minutes ago, Mateyyy said:

Roughly up to what current draw would you say is adequate for an EPS12V, or in other words, what would you consider not to be "anywhere near" that 28A mark?

 

I'm asking because the RTX A6000 takes an 8-pin EPS for power, and that's a 300W card.

240W.  Remember that the card is still capable of getting up to 75W through the slot.

 

[GoldenLag]

14 minutes ago, jonnyGURU said:

240W.  Remember that the card is still capable of getting up to 75W through the slot.

and the 4-pin is 192w, with the second 4-pin being the same in pinout, and adding only 48w? thats a significant reduction, tho i do understand it wouldnt be liniar scaling. 

15 minutes ago, jonnyGURU said:

Oh.. That was snipped from a Micro-Fit+ data sheet.  I happened to have one open while reading this thread, so just used it as a point of reference.

ah, fair enough. 

 

checking microfit 6 pin it seems like 7.5a on brass+tin and 18 AWG wire. for which i assume bronze+tin would improve that somewhat. 

[Guest]

7 minutes ago, GoldenLag said:

and the 4-pin is 192w, with the second 4-pin being the same in pinout, and adding only 48w? thats a significant reduction, tho i do understand it wouldnt be liniar scaling. 

ah, fair enough. 

Right.  Resistance + heat = resistance + heat =..... rinse and repeat.  The decay is exponential.