All RTX 5090s only have 1x 16-pin connector

[sudothelinuxwizard]

I mean, if they're gonna add more connectors they might as well use 8 pin, no? The whole point of 12VHPWR was to reduce cable clutter AFAIK. I think the ideal solution is to make a v2 version that can carry say, a thousand watts, but I guess physics prevents that. Given how much these cost they should try though.

 

[Chrispy_]

learn to read as you never mention either "12V2X6 is not 12VHPWR"

Uh, are you having a stroke?

*You* literally just quoted me mentioning exactly what you've just said I never mentioned.

I mean, if they're gonna add more connectors they might as well use 8 pin, no? The whole point of 12VHPWR was to reduce cable clutter AFAIK. I think the ideal solution is to make a v2 version that can carry say, a thousand watts, but I guess physics prevents that. Given how much these cost they should try though.

If 12V2X6 can handle 600W there's no reason why 12V3X6 couldn't do 900W, and it's not a huge stretch to imagine more pins on a single connector.

We've been using 24-pin connectors on motherboards for 23 years and counting!

 

[dirtyferret]

*You* literally just quoted me mentioning exactly what you've just said I never mentioned.

you never mentioned connector or cable just the naming scheme

I mentioned just the cable portion (hence the connecting cable is the same in bold)

if this continues, yes I may have a stroke

 

[Gameslove]

So, meltings cables in RTX 4090 is Nvidia or ATX 3.0 PSU engineering fail?

 

[Chrispy_]

you never mentioned connector or cable just the naming scheme

I mentioned the connectors only, explicitly by name.
As I said, you're confusing the connectors and the cables. The standard refers only to the connectors.

The old connector was 12VHPWR
The new connector is 12V6x2

The cable doesn't really have a name and the specification for both connector standards simply states how much current each individual wire must carry. If you search through the various power supply manufacturers, they all call their proprietary cable various things, like 12-pin, PCIe 5.0, ATX 3.0 12-pin, Type-5, and yes - even 12VHPWR on occasion - but the cable is most commonly called a PCIe 5.0 power cable, to distinguish it from previous generation 8-pin and 6-pin PCIe cables.

More importantly - you've been living under a rock if you think there's a problem with the cables. In every single thread, article, video, so far where there's been a 4090 catch fire, or melt, or start smoking, it has been the connector that's failed, not the cable.

 

[sudothelinuxwizard]

So, meltings cables in RTX 4090 is Nvidia or ATX 3.0 PSU engineering fail?

A lot of factors. Watch the GN video.

 

[Gameslove]

I mentioned the connectors only, explicitly by name.
As I said, you're confusing the connectors and the cables. The standard refers only to the connectors.

The old connector was 12VHPWR
The new connector is 12V6x2

Hopefully it comes 4 PC-E 8 Pin to 12V6x2 adaptor in the box.

 

[Chrispy_]

Hopefully it comes 4 PC-E 8 Pin to 12V6x2 adaptor in the box.

I suspect they will, since most 40-series cards still do.
Four big connectors into one tiny connector is going to look funny, but it will work.

 

[dirtyferret]

The cable doesn't really have a name

then maybe you should tell Intel that when they wrote the spec sheet, man you freaking wrong

A modular power supply supporting detachable 12V-2x6 cables, must configure the
SENSE0 and SENSE1 pins only within the power supply, for all power levels, to ensure
correct operation when interchangeable “double-ended” 12V-2x6 cable assemblies are
used.

Don't have a name? Freaking pathetic

 

[DemonicRyzen666]

Someone was saying that having two 8pin vs three 8pins was giving around +/-50mhz in gpu core mhz. Well this most likly had to do with electromagnetic induction as lower number of cabling will give off a larger field than what multiple smaller cabling will give off. Basicly the eletrocmagnetic waves are starting to interfer with clockgenerators on GPUs. I think Jayztwocent experinced this where is microphone pice was casuing instability on one of his overclocks on his cards. I personally have speard specturm enable as to not mess with my inturnal Tv tuner in my PC. Heck fan close to motherboards maybe effecting vous now & we may never notice it unless someone tests max pbo/turbo overclocks.

 

[Chrispy_]

then maybe you should tell Intel that when they wrote the spec sheet, man you freaking wrong

A modular power supply supporting detachable 12V-2x6 cables, must configure the
SENSE0 and SENSE1 pins only within the power supply, for all power levels, to ensure
correct operation when interchangeable “double-ended” 12V-2x6 cable assemblies are
used.

Don't have a name? Freaking pathetic

I love how you're getting into a strop over the non-standardised modular cabling semantics that varies wildly between different PSU manufacturers. Clearly in this case an interchangeable "double-ended" 12V-2x6 cable assembly is explicitly describing the assembly of two 12V-2x6 connectors joined by cable.

That's why the Intel spec calls it an assembly and not a cable, because the cable isn't part of the spec, only how the connectors are used an what the pinouts of the connectors should be.

Why are you so focused on the cables anyway, zero issues have been reported with cables - it's always been connectors (and the connector ends of adapter boards and adapter cables) that have failed.

 

[dirtyferret]

I love how you're getting into a strop over the non-standardised modular cabling semantics

the Intel spec calls it an assembly and not a cable,

I just quoted their spec sheet verbatim and now you are arguing with what they wrote? PATHETIC! and I love how you got caught talking out of your azz and tried to change the subject as not to look foolish. You were WRONG about the cables. I'm done embarrassing you. ATX specs calls them 12V-2x6 cables in their spec sheet, end of story.

 

[Kapone33]

If this was a problem 5xxx series would have been different.

Have you known what Nvidia is like?

 

[evernessince]

I ran my 4090 at 600w for two weeks straight trying to burn the connector... I didn't have any luck... actual wattage was around 580w it's hard to get it to exactly 600w.

Connector failure rates are low and are not going to be easy to reproduce.

Ultimately a 1% increase is monumental because in this case you are talking about a fire hazard and damage to the card.

The 5090 should have 2 connectors with how close it is getting to the max spec of the cable. All this means is that small defect or a combination of non-ideal operating environment factors are more likely to lead to a failure.

By how much we will have to see. What I don't get is the people arguing against a superior product at a detriment to themselves. You as a user loose nothing if the card had 2 connectors, there's no logical reason to defend the decision other to cover for Nvidia's Apple like nonsense.

 

[oxrufiioxo]

Connector failure rates are low and are not going to be easy to reproduce.

Ultimately a 1% increase is monumental because in this case you are talking about a fire hazard and damage to the card.

The 5090 should have 2 connectors with how close it is getting to the max spec of the cable. All this means is that small defect or a combination of non-ideal operating environment factors are more likely to lead to a failure.

By how much we will have to see. What I don't get is the people arguing against a superior product at a detriment to themselves. You as a user loose nothing if the card had 2 connectors, there's no logical reason to defend the decision other to cover for Nvidia's Apple like nonsense.

I say that, but I'm in full agreement the connector needs to be idiot proof or have a warning that you must plug it in correctly with a color picture on how to do it properly.

I personally like the new connector and don't want to go back to the 3x 8pin.

 

[ShrimpBrime]

I just quoted their spec sheet verbatim and now you are arguing with what they wrote? PATHETIC! and I love how you got caught talking out of your azz and tried to change the subject as not to look foolish. You were WRONG about the cables. I'm done embarrassing you. ATX specs calls them 12V-2x6 cables in their spec sheet, end of story.

I love how you're getting into a strop over the non-standardised modular cabling semantics that varies wildly between different PSU manufacturers. Clearly in this case an interchangeable "double-ended" 12V-2x6 cable assembly is explicitly describing the assembly of two 12V-2x6 connectors joined by cable.

That's why the Intel spec calls it an assembly and not a cable, because the cable isn't part of the spec, only how the connectors are used an what the pinouts of the connectors should be.

Why are you so focused on the cables anyway, zero issues have been reported with cables - it's always been connectors (and the connector ends of adapter boards and adapter cables) that have failed.

Oh yeah?? Both you guys are fkn wrong man.

The connector on the card is called a plug.
The pig tail is a jack.
Deutsch connectors.

12v 2x6 cables in-between pigtail ends. Its not really that specific, cause that's just what it is, just some wire. Lol.

Put that in a pipe and smoke it!!!!

 

[Chrispy_]

I just quoted their spec sheet verbatim and now you are arguing with what they wrote? PATHETIC! and I love how you got caught talking out of your azz and tried to change the subject as not to look foolish. You were WRONG about the cables. I'm done embarrassing you. ATX specs calls them 12V-2x6 cables in their spec sheet, end of story.

I still don't understand the point you're trying to make. You've contradicted yourself numerous times and seem obsessed about the cables when the cables aren't the problem, I never mentioned cables - that's something you brought to this discussion and cannot drop and neither does the spec which you are quoting. The closest thing to a cable standard is Intel denoting two 12V-6x2 connectors joined together to form a reversible assembly, with the caveats that this only applies to modular PSUs that use the standard connector for their modular backplane.

As far as I'm concerned, you must indeed be having a stroke. Nobody else cares, if you think you've won then "well done"? Let's leave you to take this win, bask in victory and get back on topic.

 

[dirtyferret]

I still don't understand the point you're trying to make.

thats a you problem that will need a better education on how ha PC hardware works to fix

. You've contradicted yourself numerous times

not once have I contradicted myself and Ive only spoken about cables hence my initial post in bold

. The closest thing to a cable standard is Intel denoting two 12V-6x2 connectors

100% wrong as usual for you,if you have never even read he spec sheet you can't talk about it and you proven you never read it multiple times. I've posted from it verbatim which it seems both the definition of the word much like PC hardware knowledge has gone completely over your head

I'm done embarrassing and this pathetic attempt to prove you have a clue to what I'm talking about. You have my permission to continue with your hissy fit and posting your drivel on these message boards which you have done since you joined.

 

[Chrispy_]

Ive only spoken about cables hence my initial post in bold

That's what makes no sense. You've jumped down this combative rabbit hole since post #17 with your hard-on for cables, but:

• This isn't a thread about cables.
• I didn't mention cables in the post you took issue with and quoted me on.
• None of the 4090 melted connectors have ever been caused by the cables.
• The cables are unchanged between 12VHPWR and 12V-6X2, so it's not even f*cking relevant here.
• The spec you cited (that I haven't read, nor do I want to read) seems to only mention connector assemblies, for when a 12V-6X2 connector is used at both ends of a cable.

We get it, something about PCIe 5.0 power cables really grinds your gears, and you'll die on that hill - but this is the wrong thread. You haven't explained once why you ever quoted me on cables, why you think I'm your opponent in The Great 50-Series Cable War of 2025, or why you keep arguing about cables in a thread that isn't about cables in the first place.

 

[OolongGoat]

Maybe I shouldn’t have pointed out the connection difference between SKUs…. I still can’t believe this is such a heated topic. It comes down to user error and ensuring the cable is plugged in properly, giving it enough room for bend clearance.

 

[chrcoluk]

Someone with access needs to ask Nvidia why they didnt use angled connectors like they on the 3080 and 3090, no bend constraints then. I get that the new revised connectors improve a lot, but still better safe than sorry.

 

[R-T-B]

I don't see how that connector can handle that when the connector already had problems handling much less power draw from RTX 4090's at 450 watts. Are they now rating by maximum transient spikes instead?

The new connector revision is fine and the internet has not caught up is how.

ATX specs calls them 12V-2x6 cables in their spec sheet, end of story.

and guess what guys? No one cares.

Your whole discussion is like this redditor that was riding me the other day for giving a brand opinion because it didn't "have citations beyond personal ancedotes"

Sure, technically he's right but nobody cared. Move on.

 

[Enterprise24]

I've seen some video talking about 5090 will need 720w of power to not being power limited (He calculated from 5080 360w x2).
IIRC the only 4090 that came with 2x16 pins was the Galax HOF.

 

[Onasi]

I've seen some video talking about 5090 will need 720w of power to not being power limited (He calculated from 5080 360w x2).

I am not sure how he even arrived at that conclusion. NV cites 575 as the TBP. What it will actually pull is anyone’s guess. But just taking double the 5080 as some target for “not being PLed” is completely arbitrary. Seems like pulling numbers out of thin air.

 

[N/A]

I am not sure how he even arrived at that conclusion. NV cites 575 as the TBP. What it will actually pull is anyone’s guess. But just taking double the 5080 as some target for “not being PLed” is completely arbitrary. Seems like pulling numbers out of thin air.

NV also cites 2.0 GHz base clock, average with RT could be 2.4-6 and far from ideal. 3.0 should be the minimum on the N4P node.