I'm not sure this is quite right, though I get where it's coming from. Unless I'm missing something, ATX3.1 PSUs, which have been around for a year or so now, might be ok. ATX3.0 PSUs might be a bit dodgier. Corsair have a bit of a corporate-sanitised explanation up.
-
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
- Thread starter scubes
- Start date
Associate
- Joined
- 8 Nov 2012
- Posts
- 399
I think its PCI-SIG that sets out the rules for what can be labelled as H+ and H++ (which in the case of the cables is just chronological - if your previously H+ cable is made after the date of the new spec, you can then put H++ on it).
Specifications | PCI-SIG
pcisig.com
Regretfully you can’t (legitimately) access those specs without a very pricey membership…!
If JonnyG is to be believed, there is technically no difference between the cables as far as PCI-SIG is concerned. If it can be used with 12VHPWR, it can be used with 12V-2x6. The changes to H++ only apply to the connector, not the cable.
I don't have a clue, all I'm going off is what he said in the article, guy knows his stuff.
I think the point - and I may well be wrong here - is that having a 12v-2x6 on the GPU end but a 12vhpwr on the PSU end is risky, as if the PSU-end cable isn't seated perfectly, you're going to get those horrifying temperatures.
If you have a 12v-2x6 at both ends, which you will have with an ATX3.1 PSU (but not an ATX3.0), the risk is reduced.
Associate
- Joined
- 24 May 2015
- Posts
- 744
Wouldn't using the supplied 8pin splitter offer more redundancy for error than a 12vhpwr cable?
Not unless you have a power supply with multiple power rails. Even then that might not help, best case in the event of an overload the PSU goes into over-current protection and shuts down, but more likely if the power rail is strong enough it will probably still melt the connector.
The problem is that these new GPU's have such large transient power requirements that you pretty much need a PSU capable of delivering enough power to melt the cable/connector.
Long term the only real solution is to load balance each of the conductors in the 12vHPWR/12v 2x6 cable, that can either be done at the GPU side (as it was for the 30 series) or in theory it can be done on the PSU side (but no MFR has ever implemented 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.
Last edited:
How so? It all joins to become one pool of 12v anyway. Redundancy would be multiple seperate cable that plug into their own PSU and GPU connector; I know it's a strange concept but think about it, some day it's going take off
Last edited:
Soldato
- Joined
- 16 Aug 2009
- Posts
- 8,109
What in the name of... why in the name of all thats holy did Nvidia decide to reduce the load balancing shunt resistors down from 3 to just 1, where its possible 600W can potentially be routed through a single wire and nothing be flagged as wrong by the card? Its mind blowing. The only possible conclusion its possible to come to is that they cut down on the power regulation circuitry for the simply expendient of saving a few quid on components and increasing their profit margins some more %. I can't even. This is bloody typical Nvidia though they removed the power regulation from the card and probably expect the PSU manufacturer to pick up the slack and supply the circuitry instead at extra cost to them and not to Nvidia. Meanwhile it means that if the the plug isn't plugged in properly or fully or if for some unforeseen reason all but one of the wires become broken or otherwise unable to supply current its possible for 600W to be pushed through a single wire and and if the wire gauge isn't sufficiently thick enough for it to actually catch fire. Unbelievable.
Last edited:
saw this on a French website
*image removed*
You better watch out for the mods. I posted a similar cropped image from a non UK retailer, with no store info at all, showing placeholders prices for the RX 9070 and got slapped with a post delete and a snarky message.
Last edited:
Soldato
- Joined
- 19 Oct 2010
- Posts
- 2,977
Luckily you've quoted the image so it will remain even if the OP edits.
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.
Last edited:
Has there been more than that one person with third party cables melting?
Last edited:
Some sort of clarity from Corsair:
www.corsair.com
Native 12V-2x6 vs 8-pin to 12V-2x6 - what is the difference? | CORSAIR
The difference is purely in the connector on the power supply, but let us take a look at the different ways these can connect to your PSU.
www.corsair.com
Soldato
- Joined
- 4 May 2006
- Posts
- 3,323
- Location
- Manchester
Two 5090s and a 5080 so far.
There's a table on this page. I'd be nervous if I had a Loki PSU...
@Nitefly Don't take this the wrong way and report me or anything but i have to ask, are you unintentionally perhaps conflating the cable and the connector?
The reason i ask is IDK about anyone else but when someone is talking about the specifics of cable/connector i think of the wires as being the cable and the connector being the bits on the end that, well, connect the wires to things.
Is this maybe where past confusion has come in? Because like i said there's very little, other than the diameter and *composition, of a wire that makes one different from another in terms of electrical/thermal conductivity.
*That's pretty obvious like solid or stranded core and Al vs Cu.
The reason i ask is IDK about anyone else but when someone is talking about the specifics of cable/connector i think of the wires as being the cable and the connector being the bits on the end that, well, connect the wires to things.
Is this maybe where past confusion has come in? Because like i said there's very little, other than the diameter and *composition, of a wire that makes one different from another in terms of electrical/thermal conductivity.
*That's pretty obvious like solid or stranded core and Al vs Cu.
Last edited:
Soldato
- Joined
- 19 Oct 2010
- Posts
- 2,977
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?
"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?
I need a new Graphics Card for FS2024 to go with my 13600K and 64GB RAM.
I'm willing to spend around £500, and want to stay with nvidia for ray tracing and DLSS4.
Is it better to wait to see what the prices for the 5070 will be? I'm thinking that they might be more than £500 and probably around £600-700? If so probably better to just buy a 4070 or 4070 super?
I'm willing to spend around £500, and want to stay with nvidia for ray tracing and DLSS4.
Is it better to wait to see what the prices for the 5070 will be? I'm thinking that they might be more than £500 and probably around £600-700? If so probably better to just buy a 4070 or 4070 super?