• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA RTX 50 SERIES - Technical/General Discussion

This is the good bit.

"Note that the above is true for the additional individual 12V-2x6 cables that CORSAIR makes. However, some of the lower-wattage ATX 3.1 PSUs we make, like the RM650e come with a 12V-2x6 cable that is only capable of delivering up to 450W."

How do you know what cable you have?
It's not the cable that tells/defines how much power a PSU can deliver from the 12V-2x6 socket on the PSU, the PSU tells the GPU via the sense pins by not grounding all the sense pins on its side.

The sense pins are not there to protect the cable, they're there to protect the PSU. Or more accurately to tell the GPU not to draw too much power so as to be constantly tripping the PSU's internal protections.
 
This is the good bit.

"Note that the above is true for the additional individual 12V-2x6 cables that CORSAIR makes. However, some of the lower-wattage ATX 3.1 PSUs we make, like the RM650e come with a 12V-2x6 cable that is only capable of delivering up to 450W."

How do you know what cable you have?
The 600W cables have a lable on the GPU end saying 600W, I assume that the 450W cables will similarly have a lable saying 450W on the GPU end.
 
I think it’s ridiculous that cards are burning up on the connection generation after generation. Nvidia need to either specify a list of approved PSUs or rework cards to run older standards that are proven and reliable.
 
This is the good bit.

"Note that the above is true for the additional individual 12V-2x6 cables that CORSAIR makes. However, some of the lower-wattage ATX 3.1 PSUs we make, like the RM650e come with a 12V-2x6 cable that is only capable of delivering up to 450W."

How do you know what cable you have?
At a guess there'll be a label on the cable? But even if there isn't, I'm not sure it matters. The only GPU that needs more than 450W is the 5090 (oh, ok, and maybe some fancy 4090 models). If you trying to run a 5090 off a 650W PSU, then I can only assume that your CPU is a potato and maybe you're using a couple of sticks of celery for RAM.
 
  • Haha
Reactions: ne0
Last 12 hours there has been 3 more burnt up rtx5090 connectors on Reddit

Yet the cards sell out day one to the beta testers that buy them. Maybe it would be wise to wait for reviews and reliability to come in before spending money on one!
 
Reading that Reddit thread just makes it sound like user error, using cables not approved for the PSU, using a very old PSU and using a cable that predates the 50 series (which the brand have since released an updated cable)
 
Last edited:
Last 12 hours there has been 3 more burnt up rtx5090 connectors on Reddit

I’ve just seen one on Reddit where the guy wired the cables incorrectly into the PSU, rather than used the “wrong” cable.

Daisy-chaining extensions too!

No harm done to the GPU or connector but the PSU and extensions were melted, as to be expected.

User error for sure, at least in that instance.
 
Reading that Reddit thread just makes it sound like user error, using cables not approved for the PSU, using a very old PSU and using a cable that predates the 50 series (which the brand have since released an updated cable)
It might all ultimately be 'user error' in these cases, but the I feel like user error should probably result in your PC refusing to boot, not catching fire :p
 
I’ve just seen one on Reddit where the guy wired the cables incorrectly into the PSU, rather than used the “wrong” cable.

Daisy-chaining extensions too!

No harm done to the GPU or connector but the PSU and extensions were melted, as to be expected.

User error for sure, at least in that instance.
Exactly.
 
Maybe, but you’re discounting FOMO.

I never quite understand that as the stock will replenish anyway, a lot of people just seem overly feverish to buy things on day one for some reason I don't really understand. If you're getting a 5090 chances are you've already got a capable graphics card anyway so why not wait for the reviews and reliability to come in first before giving yourself the mental gymnastics which comes with burning gpu cables etc.
 
Yes - that’s right. The only point I’m making is that it’s not ‘wrong’ to label or market the cable as H++ if it’s compliant with the H++ spec.

It’s also not wrong to say the cable is ‘new’ or ‘different’ to a previous version of the cable sold if there are any changes to it. If the changes are for safety reasons (such as making the ends of the cable a different colour) then it becomes a ‘new, different and safer H++’ cable.

I would agree though that some manufacturers are exploiting the situation by marketing cables as if it’s a change in the H++ spec that makes the cable itself new, hence they are deserving or ire.

The big HOWEVER though is this:



Spot on.

If you are using a PSU which is H+ then unfortunately you aren’t going to be 100% benefiting from the H++ socket on the GPU.

A H+ socket is less safe. If you connect a 50 series GPU to a H+ GPU then you are not covering the GPU being exposed to a faulty H+ connection. You're still in the ‘4090 melty cables’ zone. The H++ on the GPU end isn’t going help.

I would not be connecting a 5090 to an outdated native H+ socket on the PSU. One one hand, it’s even riskier than it was at the launch of the 4090 because the 5090 draws more power. But at least you’d be reducing the location of a ‘bad connection’ by half.

*******

TLDR:

You will miss out on having ‘all of the benefit’ from the new H++ safety standards (which aim to reduce the chance of ‘melty cables’) if you use a now outdated H+ socket on the PSU. This is particularly relevant for 5090s, which are the riskiest GPUs because of the high power draw.

This applies regardless of whether you use a cable that is marketed or marked as ‘12VHPWR / H+’ or ‘12V-2x6 / H++’.

H++ at the GPU end does not save your GPU from a H+ connection failure at the PSU end.
Listening to Buildzoid, it seems like the general consensus is that the sense pin changes on the new connectors basically achieve nothing. The issue is that the GPUs can't load balance or sense current irregularities. 3090tis had the regular old 12VHPWR connectors, drew a lot of power and they didn't have issues because they were load balanced properly. 40 and 50 series cards (at least the high power 90 series) are inherently faulty.
 
Last edited:
Some sort of clarity from Corsair:


Good to see!

It’s a shame that no manufacturer nor Nvidia are going as far to point out that the initial PSUs with H+ connectors are outdated… which I understand because it’s a ‘bad look’.

What’s going to cause an uproar is if we get a melty 5090 cable and a GPU manufacturer turns around as says:

“your PSU was H+ and you knowingly didn’t use a H++ PSU, we won’t take responsibility for your damaged 50 series GPU”.

^^^ If the GPU maker’s manual and advertising said “you must use a PSU compliant with the H++ standard if you’re using a native cable” then to be fair, they would be reasonable in saying this.

It’s everyone’s responsibility to be clear when there is safety risk - the PSU makers need to step up. And yes, it’s also Nvidia / intel’s fault for having the design flaw in the first place.
 
Apparently with the 5090/5080, the only setup that can do per wire current balancing is a full Asus setup. You need to buy an Asus Astral card and an Asus Thor III PSU. With this combo of GPU and PSU, the PSU will detect and balance power on each wire to the GPU and prevent any overheating

There is no other setup on the market today that can do this
 
Last edited:
If you have 4 separate 8-pin plugs (or whatever plugs on the PSU side that correspond to the 8-pin connector at the GPU end) at the PSU side, then you would have more redundancy up to the point it all comes together at the 12VHPWR connector itself (on the adapter).

More cables, more connectors etc.

The adapter adds some possible failure points with the four 8-pin connectors, but those connectors have a safety margin such that I have not seen a single instance of the 8-pin side of the adapters failing.
Yes this. Better heat disipation also for one end of the cable.
 
Hopefully this is helpful to those still concerned:

https://knowledge.seasonic.com/article/79-comparison-atx-3-0-vs-atx-3-1-standards

I purchased a Vertex model in Summer 2023, which arrived with the 12VHPWR cable.

There are 2 developmental difference from ATX3.0 and ATX3.1:

i) change of cable: 12VHPWR to 12V-2x6
ii) voltage power off retention time: 17ms to 12ms

The change of the cable is simply the connector pins.

Ironically, the earlier ATX3.0 models have a superior voltage power of retention!

I am happily relying on the ATX3.0 psu, and have the option of purchasing the updated cable, should I feel incapable (nervous?) of properly inserting the cable connectors correctly.
 
I never quite understand that as the stock will replenish anyway, a lot of people just seem overly feverish to buy things on day one for some reason I don't really understand. If you're getting a 5090 chances are you've already got a capable graphics card anyway so why not wait for the reviews and reliability to come in first before giving yourself the mental gymnastics which comes with burning gpu cables etc.
I couldn’t agree more.
 
Back
Top Bottom