• 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.

*** The AMD RDNA 4 Rumour Mill ***

First 9070XT victim


I see some discolouration but i don't see any melting.... this looks a bit desperate in the tech press to find an AMD example, i don't think this is it....

The Pictures of the "Melted" connector.

ivLoF51.jpeg
qREhlJ9.jpeg
 
Last edited:
Ah man, I just said the other day in a post that there had been no failures on AMD cards. I jinxed it! :cry: I was looking forward to using that connector for my next GPU, might have to re-think that now. Be interesting to see what UDNA ships with. Is there also any truth in what people are saying about the under spec'd PSU? I'd have initially thought an underpowered PSU would just crap out and shut down etc. Not cause a connector to melt. But then people are talking about voltage drops, resistance, etc. So yeah I have no idea. Is this just the one failure on an AMD card since their release back in March?
 
I thought it was caused by the cards drawing more amps from one pin instead of spreading it out, wattage wouldn't change that (?)
Also saying that to make me feel better with my 650W :D
 
Ah man, I just said the other day in a post that there had been no failures on AMD cards. I jinxed it! :cry: I was looking forward to using that connector for my next GPU, might have to re-think that now. Be interesting to see what UDNA ships with. Is there also any truth in what people are saying about the under spec'd PSU? I'd have initially thought an underpowered PSU would just crap out and shut down etc. Not cause a connector to melt. But then people are talking about voltage drops, resistance, etc. So yeah I have no idea. Is this just the one failure on an AMD card since their release back in March?

So far as we know yes this is the first...

He was using third party extension cables which are known to exacerbate the potential of the problem but i don't see that as an excuse, only a reason, it really shouldn't be happening at all.

Using extensions adds a lot more copper between the power in and the power source, (PSU to card) that does create more resistance which the PSU may up the amperage to overcome, that aside from also not being ideal for the GPU's power stages as its unclean power also heats up the points were you have connections in the supply lines.

Generally one should only use extensions that replace the original PSU cables and they should be of the same or shorter length, as well as high quality.
 
Last edited:
To expand on it using third party extensions improperly is like turning your high quality Gold Rated PSU in to a cheap, stripped out no name piece of crap. The balance between Amperage and Voltage will be all over the place, bad voltage ripple, aside from the overheating problems even if you don't have any issues resulting from that your GPU may suffer both in the short and long term as the unstable voltage its going affect the cards stability and the dirty power puts unnecessary stress on the voltage regulation modules prematurely wearing them.

That's not to say these cables are a bad idea, they aren't, they look nice, they can be of a higher quality than your original cables and you can use them to get a bit more length, but not a lot more and again use them instead of the original cables, not in addition to.
 
Last edited:
yes think some psu manufacturers are introducing safety features on new psus to monitor each wire so if the load isn't balance on all cables it will shut down

These GPU's could and IMO should have the same system, but that requires the GPU end to have separate intake circuits, currently those connectors while appearing to have 6 pins top and bottom are internally just solid metal plates creating just one connecting point with live and ground only being separate, it might as well be one fat cable, that would actually solve the problem too.

On top of that you would need balance monitoring hardware and software, all this would add cost which would defeat the objective on the cost savings you get from just having one connector instead of two or three.

Edit: The old 8 pin connectors are separate pins. These 12v HP connectors are cheap trash. Like how the Chinese take a high quality design that works and figure out how to make it cheaper and crappier Chineseium.
 
Last edited:
Is 16AWG wire better than what most other OEMs are shipping? Are OEMs using 18AWG for example? I like the idea of this as it is a thermistor directly connected to the cable so will detect any heat build up and cause the PSU to do some proactive protection (I guess).

 
Last edited:
Bleh. This thread has just prompted me to check my own Taichi, used since Saturday after UK release.

It's fine. I should have known better.

For reference, I'm using a 2021 Corsair RM1000x PSU with the extension cables that came with the card.
 
Last edited:
Basically Nvidia has way more driver draw call overhead than Amd. The way this manifests itself is that any game that hits your CPU hard will result in Nvidia GPU losing some performance because the CPU is too busy with the game to fix the overhead problem, thus allowing amd GPU to close the gap
So basically Nvidia offloads processing to the CPU to cut down on component cost in the GPU (and increase profit margins in the process). Why am I not surprised.
 
The issue is about supply. You can barely find AMD in pre-built systems and laptops - Nvidia holds a monopoly in that market. Just because DIY sales might be OK by AMD standards,isn't much when they are fraction of all dGPUs sold. AMD doesn't seem to seriously be addressing that market - they could always have used Samsung as a second source,but weirdly never bothered.

They are literally selling fire hazards and have increased their market share. But if people get away with things, they will keep on doing it won't they.

Because most dGPUs are sold are in pre-built systems and laptops. A lot of those will be derived from the dGPUs in the 60 and 70 series desktop cards. They work fine. Navi 44 is nowhere to be seen in laptops and is rare in desktops from larger companies such as Dell,Lenovo,etc. Strix Halo is priced way too high - unless you need a lightweight laptop,an RTX4060 or RTX5060 laptop is cheaper.

In the UK at least,it's easier and usually cheaper to find a system with a Nvidia dGPU compared to an AMD one. I can find an RTX5070 based desktops for under £1000.

But the UK is one of the easier markets to find AMD hardware - now think of all the other markets where AMD barely supplies hardware?
 
Last edited:
In the UK at least,it's easier and usually cheaper to find a system with a Nvidia dGPU compared to an AMD one. I can find an RTX5070 based desktops for under £1000.
I think this is such a key point, particularly when both the 60/70 class is the real 'main group' for PC players - and then prebuilts are such a popular route now too.

It's so much easier to buy a pre-built 'affordable' PC with an Nvidia GPU with a 60/70 tier class of card than it is to find an AMD equivalent. If I look at a few of the more 'popular' pre-built companies in the UK, the top 8 out of 10 best sellers on one in particular are all prebuilts with Nvidia cards.
 
Back
Top Bottom