The 12VHPWR adapter cable is DANGEROUS! But NVIDIA doesn't agree...
HTML-код
- Опубликовано: 14 окт 2024
- Get your JayzTwoCents Merch Here! - www.jayztwocen...
○○○○○○ Items featured in this video available at Amazon ○○○○○○
► Amazon US - bit.ly/1meybOF
► Amazon UK - amzn.to/Zx813L
► Amazon Canada - amzn.to/1tl6vc6
••• Follow me on your favorite Social Media! •••
Facebook: / jayztwocents
Twitter: / jayztwocents
Instagram: / jayztwocents
SUBSCRIBE! bit.ly/sub2Jayz... 
Наука
Test the cable you bent, and use a Thermal Imaging Camera to see if it gets hot (or even melts).
Yeah, I pretty much already resigned to the fact that Im going to have to do this.
@@Jayztwocents You will have my view! And my like!
@@Jayztwocents waiting for the vid now. Be safe a fire extinguisher handy! You maniac haha
These cards should come bundled with a high quality surge protector now. 600w through these tiny gauge cables…… add to that a connector that was designed and built with anything but safety in mind…….
Or just wait a few days for more cases of this
Even with all of the red flags, Nvidia's refusal to accept that this is a legitimate problem has always been baffling to me
it would cost them money. money is more important than customers
@@pagannova3621 Aren't they already losing money anyways?
It’s just like car companies. Until the cost of settlements becomes higher than the cost of recall, people can die for all they care.
Even with all the red flags, people will still continue to buy from them. Look at when it was announced about that amazon MMO wrecking some 3090's I think it was... did it stop people from buying the game after the face 'just to see'. Nope
Just like how the 4080 12g was "Confusing" people. Kinda feels like there blaming us for them trying to rip people off.
Intel not using the very plug that they helped design is absolutely hilarious
It wasn't ready honestly for mass use. Besides their graphics cards aren't using power too much
Giving me Apple and USB C vibes.
It really cracks me up with this theory. 🤣
their cards don't need anywhere 600 watts
😂😂😂😂
The more problems there are, the more the EVGA breakup seemed to be a 200iq move. Dude if NVIDIA has to spend money redesigning every 4090, EVGA is going to be LAUGHING at them. Oh my god this is too good.
They'll probably just make a new adapter with a right angle plug.
All they have to do is not use a plastic that’s in a ton of computer parts that starts deforming at only 105C. ABS plastic shouldn’t be in computers anymore
Manufacturers should have freedom of choice in power connectors
I mean, I have yet to see a report of this happening on Founders Edition cards. So far all the legitimate sources say it’s all reports from AIB cards. Which include non-Nvidia cables.
Also, EVGA isn’t going to be laughing anytime soon, their mobos aren’t really relevant and that’d be a hard game to break into anyways with MSI and Asus in the mid and high end with Asrock (literally just Asus) covering the budget end, and while their PSUs are good, there’s better ones at similar price points usually (for example Corsair, Seasonic, or Asus’ ROG strix line). Their AIO coolers are also very mediocre. Like…what are they going to make that’s relevant to keep the company alive? Switching to AMD in GPUs would’ve increased their chances but it’s not like RDNA is relevant in any meaningful way.
Weird comment.
Hey Jay, great video and we had to expect this, I have been working with 12V DC circuits for the last 17 years, formed as a car mechanic and now operating the service department of a Semi-Trailer dealer and I simply can't understand how the PCI specification can rely on these tiny connectors for powering 600W, that's 50 AMPS at 12V! You should look at the specification sheet of Anderson Powerpole PP15/45 (I looked at document 2020-0055 DS-PP1545 REV C7). Those connectors are rated for a maximum of 45 amps and are WAY stronger than what PCI/Nvidia uses. Also, those ratings are with 10 AWG wires, that's 6MM (or 1/4") of copper! The PC industry does not seem to take into account for "what can go wrong will go wrong"...
powerpoles was my thought as well.
I would just like to gloat, I predicted serious safety problems the moment that 600W spec leaked.
Totally agree bud.
But that’s where multiple pins are supposed to come into play being 50a/12 pins is a bit over 4A per pin.
Now if they’re not spreading the load across all of the pins properly you’re going to have issues but a single 16 awg conductor can take 13 amps on its own so on paper it seems doable if the connector is rated for 8.5A per pin.
Stress points on crimps like he points out is definitely something that can come into play and he’s not wrong in thinking that as a starting point IMO
@@Mr.Morden Who are you gloating to?
Cablemod now has a 90 degree adapter for 12VHPWR. I have a feeling those are going to be very popular. They should really come in the box for any card using them.
VHPWR = Very Hot Power
why is it 12 hp? 786 watts is 1 horsepower. Are you saying that all 12 leads have full power going through them? I didnt know that. Thats a shitload of electricity. My houses AC has only a half to 3/4 horsepower. Thats Krazy!
It's Voltage High Power
@@richiebricker Um, 12VHPWR -> 12*Volts* High PoWeR?
It's so high it can see tomorrow
Putting that much power through such a small connector immediately made me think that’s gonna be a problem. Can’t wait to see the prebuilt thermal volcanos go up spectacularly.
I think they should have looked into RC model parts for some suitable connectors.
The 3090ti variant of those connector had some genius because of the slightly put inwards tendencie... but on the 4090?... Time for "hard"wire-modding the splitters on the pcb of the card!
When I used to race 1/10 touring on a xxxs losi G+ I was running 6 sub c 3300mah nimhs that we had soldered together with bars made from melted down 24k jewelry and we soldered our leads from the speed controller to the battery packs every race. No connectors back in 01-02 for me.
Honestly agreed, even doing 500w clean amplifiers in cars don’t use thin cables, so to me using those tiny pins is baffling
Those molex terminals, if quality parts, are perfectly fine at those currents. So the concept, on paper, is fine. The issue is that they have a lot less wiggle room than the old connectors - mainly because the ATX spec had a 2x safety margin. So less than perfect (read cheap parts) still wouldn't melt stuff.
8.5 per pin would equate to around 600W (612W to be exact) That's pretty close to the spec, maybe not much headroom there... (Before anyone asks how I worked that out, it's 8.5A x 6 pins (because the current path flows in through 6 and out through the other 6, so that's how many you count) x 12V (Watts = Amps x Volts) So 6 pins * 8.5 Amps = 51 Amps total, multiply that by 12 Volts and you've got your 612 Watts. The problem I think is that they're running the connector so close to it's limit that unless ALL 12 PINS make a really good contact, the current will be shared unevenly and may put extra load on the other pins both heating the pins that don't contact properly due to resistive heating and also overloading the others that take the additional load.
Best way to use 4090 is to just decrease power target to 80, significant boost to efficiency and eliminates majority of the risk from connector.
I think connector is great when used at or below 450, problems arise when it is being pushed to 600.
@@erenbalatkan5945 but then it is better to stay on older 30xx series and push it to the max and having nothing to worry about lol. seems very counter intuitive
@@erenbalatkan5945 Under normal usage the 4090 rarely ever exceeds 450 watts anyway, so this isn't much of a problem.
Tfw youtube comments sections understood Nvidia products better than any of the higher ups assume to
So basically. People are right, in that the PSU should send more volts than just 12 to the GPU. But what PSUs can do that? None of them.
I’ve seen this 100 times in the comments, with just more “Nvidia bad” fluff.
I love that an industry standard was pushed through that has such specific caveats. 100% safe and no issues...I mean, as long as you don't bend the cable within 35mm of the connector; which we realize is impossible to do in the majority of cases due to the size of the card. That's a consumer problem, though, I suppose.
That diagram from Cablemod is EXTREMELY concerning. Kind of bonkers that Nvidia is dismissing this out of hand. Do they really expect people to always have a 35mm relief before the bend? Do they not realize that many people building PCs basically just jam their cables in their cases and call it a day?
Exactly. That responsibility certainly wouldn't fall on the shoulders of the R&D team that spent years developing this pile of shit or the company that opted for cheaper connectors in lieu of a system that worked perfectly. Obviously this is the consumer's fault/issue/problem and responsibility.
Here's the thing though right, I have the ultimate solution and the tech industry isn't even ready, it's never been done before... What if we, and hear me out, used 2 connectors...
The 12vhpwr has 2x as many +12v pins as an 8 pin power cable (6v3) and 1.5x as many grounds (6v4). So what if instead of pushing twice as much power per connection we simply didn't.
Oh noo, the newly bridge is fall and eat many life in The process. Well that's the bridge user fault, i suppose
I was working in a server centre on Friday upgrading their incoming supplies and talking to the server techs they're dreading the new spec rolling into professional and server grade hardware because it's even more of a confined space and bigger costs if something goes wrong. You know a spec is flawed in any industry when the people that are doing day-to-day maintenance consider it a step backwards/ danger.
No changes ever! Just me on my cellphone eating Chik-Fil-A before I fall asleep!
When the servers catch fire because of this, and burns down the building, then they might learn the lesson.... Maybe.
@@NorseGraphic Datacenters have much better protections than any home.
I have a degree in electrical engineering, and although my area was more on the IC side of things, 600W at 12V which is 50A of current through that small of a detachable plug is just plain crazy. Think of your NEMA 14-30 Dryer plug, that is rated for 30A.
Maybe its time to raise the supply voltage in a PC now that parts are consuming so much more power. 24V or 36V will reduce thermal loads on cables and connectors a lot and still be safe voltages and the insulation requirements are not gonna be different compared to 12V, and the PSUs use DC-DC converting already so making new versions should not be too hard. There will be a period of pain of compatibility for sure but it will be worth it in the long run.
Heavy duty commercial automotive guys like me are laughing at how stupid this is too. I've seen OEMs pull this kind of stupid shit too with Halogen light circuits. Where the pins burn up in the breakout harness connectors after the copper oxidizes from sustained electrical load from years of use, increasing the overall circuit resistance. The designed circuit load rating for the gauge wire and size pins are no longer within spec, and the connector pins are the first fail point, because they're the smallest pieces of copper in the circuit.
I agree, I worked with 24vdc systems and the amp draw as you stated is alot lot less. 50a on that small of a pin would never pass in aircraft world. This is truly an expensive design mistake.
Moving to 24v or 36v is a great idea. You just make new connectors that don't fit the current stuff and problem solved. It'll start with commercial adoption and move down to retail. But the current draw now is becoming insane for the cheaply made PSU we're all being sold these days.
What I can’t believe is that an engineer said they just replaced burnt cables. I’m also EE and have designed MANY boards with high power interfaces. This connector doesn’t seem derated very well and with POOR strain relief.
Any burnt ANYTHING during verification testing led to an automatic halt to testing as it would be considered a safety issue.
Sad!!
I am also an electrical engineer. Another solution would be for everyone involved to see reason and not think 600 watts consumed by a single PC component is anything less than insane. This is the result of expecting and pushing perpetual growth in specs. We can no longer make the same gains we used to so now we dump more and more power into bigger and bigger components. We knew 10 years ago we were about to be pushing right up against the limits of physics. We couldn't get smaller and more efficient anymore, only way to go is up and out.
I put together a video editing PC at work with a 4090. That was my first time seeing that connector. The 4090 needs 4 of the traditional GPU power supply cables/connectors. Being handy with electronics I know that electric current delivery has everything to do with the gauge of copper wire. I also know that using connectors limit the amount of current that can pass through. You have to overspec the connector to handle more current than the current limit of the wire. Typically those connectors are much larger than the wire to allow more surface contact area on the pins.
It was weird seeing these 4 large power cables go into this relatively small GPU connector. And the 4090 is almost twice the power consumption of previous generations, so that makes it even more concerning.
I’m mostly confused why they cared to make the connector so small when there is plenty of room to just keep doing what they have been doing for years.
It really does make you wonder why they didn't include a 90° connector to start with. They had to know from their own testing that it was going to be a problem.
bending cables and connectors costs money so go figure
Absolutely. That seems to be the easiest and most safe solution for now, but comes at additional cost on the customer's side that's not really optional if fire hazard is a possibility.
Nvidia and the AIBs that shipped the Nvidia adaptor should send their customers a free 90° adaptor.
Also, AMD please take notes on this to avoid a potential crash course.
@@thelegendaryklobb2879 For the price of the 4090, they should send at least 3, lol. Especially those trash plugs that were rated for like 30 inserts.
@Carl Gunderson Translation for non-Americans please
What do u mean ,quote on quote "it all just works man"
Nvidia needs to ship the straight connector version, as well as 90 degree connectors in both axis (bending back towards the backplate, and bending towards the back of the card). Building the turn into the rigid connector and using it appropriately is the only safe way I can see this moving forward.
btw some cards have inverted connector so we should have two variants of 90 degree connectors for both axis. 1 straight connector, 2 connectors for cable going up for possible vertical mount (regular and inverted), 2 connectors for cable going down (regular and inverted), and possibly 2 connectors for cables going along the side of gpu (regular and inverted). I'm LOVING this future-proof design made by the most brilliant people of hardware engineering! (sarcastic claps)
I came reading the comments before suggesting this and I'll bet cable mod and others are doing this already. If not trade mark pending..... Kindjie & Davo power adaptors Pty Ltd.
You mean re-invent the 90degree connector from EVGA... they have an adjustable power rail for the 10/20 series cards...love mine used for both series over the last 5 years...
Absolutely. And they need to provide one free of charge to everyone who purchased a card with this connector on it.
@@Tink-GB EVGA makes a good one. I don't know who made it first, but I'm using some decent no-name 180º 8-pin connectors from China for my 3090s. Obviously not needed, but they look clean and reduce cable strain.
I remember using similar connectors for IDE and Molex connectors in my cramped 486 in the 90s! The technology exists, dammit!
I dunno how the "smarts" in these new connectors work, but I bet a nice black PCB with straight through runs from one connector to another would work fine.😂
Thank you Jay for putting consumer safety as top priority and not worrying about likes, dislikes, etc. This is the info that NEEDS to be shared throughout the industry and with builders and enthusiasts. Thank you for what you do.
If this was NZXT we wouldn't be hearing so much lol
who is giving him dislikes for this lmao? The likes are overwhelmingly in his favor. He isn't martyring himself over this.
@@_BangDroid_ Do you really think that if it was nzxt jay wouldn't bring up a potential fire hazard? lol. Ok.
@@_BangDroid_ He has videos about the NZXT H1 fire hazard, he reported on it the same as everyone else.
your being to kind. You raised this issue before anyone. Credit to you..keep up the good work
You're not fear mongering at all, even if that particular burnt connector, turns out to be some strange corner case. The 12VHPWR connection is new, and is essentially being beta tested right now. We're all still learning about it. So, I hope you continue to bring us updates, as the latest information about the connection unfolds.
@Optaunix lay off the weed bruh.
The card was vertical with the cable only bending sideways due to it own weight and route of the cable. It was not over flexed at all. Cable mod attached a picture showing the cable must only bend downward and at the 35mm distance from the black connector. If that's makes sense. Over clocking hardware or whatever his name is did a vid with more info and pics.
@@Ryan-re1rs so what do u think cause this to happened ?
We shouldn't be learning on customers. This is not some fringe knowledge. Dangers of passing high current through cables is well known. The science is there, they just chose to ignore it.
The connector is 12V, right? The card is rated at 450W? 450/12 and you have 37.5A of current, at 600W it's 50A. It's very hard to believe they didn't realize that before final product.
@@JorgeForge I agree with you wholeheartedly. Just to be clear though, I only mentioned that they *are* beta testing it on us, not that I agree with it or think that it's okay, by any means. It is, what it is, though.
Whenever I see higher power requirements and see the connections get smaller, I start to question things. I had a feeling, and I’m glad I wasn’t the only one who was suspicious.
Eyup, really scares me, the Type C cable
Just wait for 1000W via Qi. 😎
Yeah, its like the manufactures like Nvidia, AMD and everyone else are trying to ignore fundamental physics or at least try to sell it as a "non-problem".
But, oh wondorous wonder, it is a problem which they then want to try to sweep under the carpet or still try to tell everyone "Hey, he's just stering up fear, don't worry - buy our damned overpriced products"....
@@finestPlugins That's called an induction stove top. 😅
@@ilovefunnyamv2nd This is true, but on the other hand I have yet to see a single USB-C cable or port fail on multiple devices. Compare that with Micro/Mini USB, where the port or cable almost always fail after so many insertions and require constant wiggling to make a solid connection. The connector itself is equally as important for safe power load as the gauge of the wire, and USB-C is simply a damn good connector.
Also, remember that your fire extinguisher needs to be rated for electric fires.
smoke alarm too otherwise it's pointless!
Should fit a fire suppressant system inside the case.
ABC class car fire ext. works
Also make sure the fire extinguisher is actually in service period haha
I'm already feeling sorry for the poor sod that let's rip with a powder extinguisher. That's some next level mess to clean up after 😂
I agree with your stance on this situation. You encountered an issue and concern and you're just giving everyone a heads up and warning. Now users can be mindful about this and hopefully take care.
When did Jay encounter an issue exactly?
I wondered about these connectors back when I first saw them. I'm an auto technician with training in hybrids and electric cars and I see melted pins in plastic connectors quite often. Knowing what I know about the theories of connector design and the wattage the new 12 pin is supposed to carry, I was fairly sure they were on the limit of something that small.
Imagine running into a bent hs-can circuit from the factory 😂 or better yet, a bent hv wire assembly from the hv battery bwahahaha
@Luka Tim Pascal dropped with effort. 2000 series was a scam, 3000 was what 2000 should have been. They're scrambling and alientating >.>
Totally agree. I have some connector/wires experience and when I saw the connector I got a bit concerned. It makes me wonder just how close to the edge they were getting. When Jay put it near the case and showed how little area there was to bend I got really concerned because bending is terrible, not just for the connector but the wires in general, especially at these higher voltages.
Based on the CableMod email ( @5:10 ) and the PCI Sig slide ( @3:13 ) it looks like everyone from the sub-tier suppliers all the way up to Nvidia (and I wouldn't be surprised if some big-wigs at Nvidia were shown these issues) knew that it was a potentially massive issue. To have the nerve to send Jay that smarmy, Emoticon-ending email just reflects badly on their team, especially when they obviously had many concerns themselves internally.
I am certain Nvidia marketing department is super thrilled to see that branded melted connector.
Maybe this is karma for all the push nvidia has done in the past to push for software developers(on the 3d industry end) to use software that is nvidia specific. Hopefully nobody gets hurt in this and nobody's houses burn down, BUT hopefully this pulls nvidia's absolute monopoly in the market back a bit and causes their footing to slip so that more software is gpu agnostic and the "industry standard" no longer has to include an nvidia card to get the job done. There are entirely too many scenarios in the industry where if you don't have an nvidia card you quite simply can't use a particular program in a production workflow environment. Its been this way for a long time too. HOPEFULLY this cuts them deep enough that they have to take a back seat for a minute so that their hold lessens on the software developers who can open their workstations to be more compatible with amd and the new intel cards and any other gpu player that might enter the market. I say this because i work in the 3d graphics industry and I have ALWAYS been an amd fan long before they were amd, I always liked their architecture for their gpus and they make really great, and more importantly, really stable cards in my humble opinion. And while I do currently own two amd gpus, both of which are quite nice, I can't use them in my typical workday. I have to use my nvidia machine because i use too much software that relies on nvidia specific stuff and if I use amd I have to do the calculations with my cpu which is job suicide because it is so much slower. And the saddest part is, my amd cards are techically faster and better than my nvidia but theyy might as well be paper weights in the industry. Especially when you consider than in any given day I have to use 6 different software programs, and no less than half have an nvidia specific pipeline. (and the ones who don't often use something like opengl or similar which isn't always as up-to-date and causes its own issues)
@@bryanharrison3889 Hopefully. I'm just a gamer anyway. AmD will be my next card.
Is it at all possible that JonnyGURU was just being sarcastic here? Just a quick glance strikes me as pure sarcasm.
Ow the PR team is sending dms to people with burned connectors.
@@zkilla4611 AMD is using the same connector on their new cards. PCI-SIG's standard is the main problem.
i love that CableMod is putting out a really stout 90 degree 12hpwr connector so your cable drops back along the surface of the card instead of pushing against cases and whatnot. i feel like those are going to sell really well
Not to mention it looks way better than a spaghetti monster coming off your gpu lol
in light of this R&D debacle, those connectors should be included free with each already overpriced card.
@@bryanharrison3889 While I agree in principal, your already a whale for buying a 4090 in the first place.
@@jonbeckham6463 Or maybe some of us just saved up... I for one am still using my 970.
@@k9px retired the 390 nitro for this
Oh boy, thanks Jay and everyone for the heads up!
Going to check my 4090 right now as I bent the hell out of that connector to get it to fit. I have been some plastic smells coming from the computer since the card was put in. But it was a full rebuild with a new PSU so just thought it was stuff bedding in.
I don't want to look!
So far I have been very impressed with the card, GPU temps are normally under 60c, never seen it hit 65c. 120hz 4k gaming with everything on max. Mostly Vermintide 2 and Metro Exodus EE.
I'll report back the good or bad news, plus the computer components.
EDIT: looks like I caught it in time. The "special" power cable was bent as much as i possibly could be to fit the side window on. I had bent it under the card though as that was neater.
I checked the female connectors in the cable. The ones that had to go the longest distance, outside of the curve, had depressed in the connector around 0.25 to 0.5 mm.
I straightened the cable and the depressed connectors did bounce back a bit, but not much. I am now running with the side off the case.
Now I need a new case with vertical GPU mount and a new special power connector.
They should make 90 degree connectors that have the plastic also hold everything together for a longer distance, so when the plastic ends it is already at the point it is ok to bend.
sic, they’ll be a big seller
Cable mod has one after some quick googling for the 12 pin
While this is the right idea for the problem, its still a Solution to a problem, which didnt exist until nvidia went hurrdurr on the totally fine standard connectors.
It's stupid you have to watch so much on the bending of the cables, in an already cramped space because the card it self takes up most of the case.
Maybe the card developers should rethink there designs. No more "MoRe pOwEr = MORE POWER" and more "More efficiency = More Power" would serve us all better.
I'll never understand while every new cpu and gpu generation becomes more efficient the manufactures see the need to ramp up the power to squeeze the last 2 percent out of it to win some benchmarks.
@@CeLLeGER i mean you dont need to use the extra power, pretty sure they are more efficent than 30 series too.
@@Neishy4AGTE Though the main issue is that new CPU's and GPU's kinda fully overclocked out of the box. Sacrificing the last 5% of performance will reduce power consumption by about 30% or so. This lowered power consumption is more safe, cool, and quiet to run. Often enough performance is more consistent too if you no longer have a temperature bottleneck.
Hello,
I've worked with plenty of electronics, including plenty of high power systems. I agree with you that it is absolutely a contact issue. Despite the smaller connector, current should still not be a problem, cables of that thickness can handle well over 600 watts. However, poor contact as a result of stress can absolutely cause overheating.
So if i buy MSI 4090 the cable MUST be as straight as possible?
If you have legitimate concerns regaurding safety of power connectors, it's better to say now and help prevent a serious accident. You done good here Jay. keep up the good work that you do.
We are Teclab and made the tests (yoi called us Galax).
We run more than than 25, considering more than 12 minutes overload (rising 300W per step), and reached 127 Celsius.
Neither connector or cable melted, and it's still working.
Then we installed bad mounting (bad assembly, both horizontal and vertical, then temperature increasing over the time, what can cause damage and melt.
About 150 min testing alive.
Thanks
You’re 100% right, that is a connection failure issue. If that was a too much current/heat problem, that card would have melted down and suffered a way worse “catastrophic” failure.
Which, according to sources for channels like Moore's Law Is Dead, was the case when they were testing a 40-series Titan prototype in the Nvidia labs, going with a 600W+ firmware profile. On some occasions, the connector would melt, maybe ignite. On others, the card itself would catch fire. Needless to say, there won't be a Titan GPU this generation for the foreseeable future.
I should add that it's why the 4090 is limited to 450W right out of the box. One could experiment with overclocking by making sure that all 8-pin slots are occupied and feeding power so that they have the 33% voltage headroom.
@@SelecaoOfMidas It's honestly pathetic that GPU manufacturers push power limits to boost performence...
PCIe spec should have power consumption standards like 5, 15, 30, 70, 150W and so on that would need to be followed.
This would force companies to truly innovate their designs, instead of just nearly frying every card and home breaker just to squeeze some FPS...
@@shapelessed I don't think we should've ever exceeded 200-250 watts at most. I was fine with 150 watts really.
Damn, good job EVGA seeing the problem, being told not to worry about it, and telling Nvidia that enough is enough.
I was thinking the same thing. EVGA abandoned ship at the perfect time. I truly hope that they will join up with AMD. Might not be this gen, but maybe next gen. I know they said they don’t plan to do so. But things can change. I hope to see them selling GPUs again.
@BainWrangler Might wanna get that RMA’d. I’ve only ever used EVGA PSUs and I’ve never had any issues. Could be a bad unit you have. They have awesome customer service and most of their PSUs have a 10 year warranty so I would give them a call.
@BainWrangler I have always used EVGA P2/G3 PSU's 1k+ watt units and never had a problem at all with coil whine at all its incredibly quiet and well within your warranty to RMA it.
@BainWrangler Is coil whine the same as cap whine? Hot glue is often used to tame coil whine.
I think it's happened to what? Two people so far? The sky is falling...
600w at 12v is 50 amps. 50 AMPS. That’s an insane amount of power for 6 little cables to carry. (Uni direction)
That's 8.33A per pin, if all the pins carry load equally. If those pins are rated to 8.5A then potentially it could be sitting right on it's load rating for hours at a time.
18AWG wire can handle 19A per wire, so almost 120A total. The issue is the pins in the connector not making proper contact, raising the resistance and thus heat generation.
@@SixDasher They shouldn't try to make a high load connector, small and dainty and possibly easy to break, or not connect properly. The current PCIe connectors are chunky and safer.
@@SixDasher 18guage wire is rated for 14 amps at 90c. This can vary. For example, at 12v it can only handle 10amps to be at an acceptable temp. Which is too close imo to the 8.33 rating per wire/pin.
@@matthewfriemoth2502 The 12VHPWR spec requires 16AWG wires for all +12V and Gnd
JAY! If you bent the pins, you might be able to put the connector under some magnification to see if the pins moved in their housing. You'll need a bright light, and maybe a microscope, like Lewis has. (Maybe. You might be able to get away with a decent magnifying glass.)
It might also be a good idea to purchase a couple of cables and do before/after shots. Lot cheaper than sacrificing a 4090.
This is EXACTLY the kind of thing that starts small and slowly becomes a crisis; but it won't happen for most people during the first few months or even years. With even a nominal horizontal force on the plug, the thin metal parts in these connectors which directly contact the pins will fatigue over time and split open at the seams. It's like the Xbox 360 "ring of death" situation where they gradually self-destructed. NVIDIA needs to get ahead of this and send people proper adapters with robust, possibly angled connectors and not to just be wrapping them in thick tape. EDIT: If you own one, make sure to watch Buildzoid's video "The 12VHPWR connector sucks" on his channel Actually Hardcore Overclocking.
Yes, this!
On every contact interface between metals, corrosion will gradually increase which will increase the resistance which will increase the heat which will speed up corrosion, and unless that is taken into account with overspec, it will start to melt and burn.
Yes, the wrapping is a really bad idea. It only increases the leverage of the cable on the connector. If they keep the connector, they will likely have to supply a selection of angled connectors with highly flexible cables to reduce forces on the connector.
Yeah but the ring of death won’t burn your house ….
@@Echristoffe Oh, I just meant how it's the kind of thing that will creep up over time like with the first 360s. I edited it to make it more clear. But yeah, this could get bad if people out there aren't aware of what's going on. Hopefully the word has gotten out.
@@ThisSteveGuy seriously after seeing the second video… I don’t know … maybe a bad batch ? Maybe some weak female connector ?
It feel weird that they have pushed a standard which is only used by Nvidia … AMD and intel are not using it for their latest gen …
Don't give up Jay, this is not a joke, people can lose their homes and / or their lives using this crap. I am surprised that the FCC / FTC are not looking at it yet, maybe we need to talk to GN to push it further (they have experience in these matters).
Thanks for making people aware.
Agreed, this needs to be investigated further.
This maybe an isolated case but if it becomes an issue to several users, FCC/FTC will come in. This could be a fire hazard.
Most likely some people need to die first in a fire caused by this issue before these arrogant companies like Nvidia and Corsair will start putting SAFETY BEFORE PROFIT AND PR BULLSHIT... ☹☹☹
6:33
"The connector is fine as long as the companies card is not mounted in a case. We've tested it under normal use circumstances."
-Nvidia QA, interviewed in the Nvidia test lab, next to the gpu on a test rig.
Awesome video.....but 13:20..."AMD is using this plug"...why are we intentionally spreading misinformation?
If a pin gets hot enough to melt the housing there's a good chance the metal has changed either in temper, or plating integrity. That can cause more resistive snowballing.
Maybe bad crimping can cause it too? Still think this plug is way too small to carry up to 600w. Hell, the 8-pin cables aren't optimal, but at least they spread power a lot more...
@@GrimpakTheMook it would have to be a really bad crimp, like very little contact which should never make it past their deficiency process. Like borderline breaking off at anytime bad crimp.
Those pins aren't tempered and how they're affected depends on exactly how hot they got and the plating used. Assuming the housing is ABS and the plating is tin, you have around 30 degrees of leeway. Not a lot, but the plastic will become soft and look "melted" way before that while the plating will not budge.
Is this why EVGA dropped out of Nvidia gpus business? I heard they burned gpus before.
@@fynkozari9271 it's not just this.
The more the 40 series is discussed, the more EVGA’s decision to part ways from NVIDIA makes sense
Yep, they saw this shitstorm coming 10 miles away and got out of the way. Good for them.
Thanks for pointing it out, must’ve listened to their internal engineers.
I believe they would use 2x 8pin plus some 4pin for data, and reduction cable for it eventually
The wall of EVGA cards behind Jay is perfect!
I guess if the new 12 pin adapter is mandated from Nvidia and evga engineers calculated that it will melt the. I guess them opting out is valid since they said something along the lines of "nvidias requirements became impossible to fulfill" iirc
This PSA and concern for safety are really helpful to consumers 👍, and we cannot thank you enough. Your bravery in the face of criticism is legendary. Kindest regards, neighbours and friends.
I know it's a stretch but I mean... this video could literally save somebody's life.
I hope it doesnt come to that ig but this is crazy stuff.
I saw this on reddit by Google app earlier. This has happened before according to reports.
Actually about the bending, Igor from Igors lab said, that when you plug in the cable FIRST and then bend it (which is, of course, easier, because it is fixed on one side now) the whole pressure goes to the connector. And the pins (card side), which are quite thick, can then bend/open the female part of the connector (cable side) just a little bit, which also leads to less contact and thereby to a higher resistance (=higher temp). This and the fact, that the outer cables are only soldered to one little metal plate, which breaks off more easily than the middle cables, which are soldered to two plates at once. And the connection from the little plates to the 'female' part of the connector is REALLY thin.
Knowing most users have to bend the cable somehow anyway, the "easy fix" would be a redesign of the adaptor cable so the metal connections are stuck in a 90 degrees plastic housing. It would eliminates the "first bend" issue, make sure the pins cannot be pulled off the housing and the product would look better anyway.
A better solution is a recall and a redesign of the cards themselves to make a new standard with bigger pins or use normal 8 pins connectors. But they will not do it, since it is way too expensive. I would have many more questions on that connector testing... I hope some regulations apply here.
Or just buy a 3.0 psu
They could simply offer a free change of the connector. Like people ship their graphics cards back to NVIDIA, and other manufacturers, and then they take off the current connectors, and change it into something else.
If this isn't possible, then they can simply redesign the PCB while keeping the rest of the parts the same, and just charge people $20 or something, or do it for free and take the L
nvidia trying to beat NZXT and Gigabyte for the title of "how many PC's can we catch fire before someone notices"
You Forgot Asus Motherboards
@@Danielkn0686They at least identified the error
@@Sonyoucef 🤝
Looking forward to the GamersNexus montage
It's not just Nvidia's fault, even if it were nice.
Intel and PCIe speks are the core problem.
Nvidia has reported that there are problems.
Based on the gauge of the wire, the theoretical maximum before failure is around 57 amps. The connector is currently specced for 50 amps at 12 volts (volts x amps = watts), so there's already not a lot of margin for error, just over 1 amp per voltage carrying pin. If there's some kind of failure in the connection integrity inside the connector, on only a few pins, the resistance of those pins will rise high enough to force more power through the pins that are connected properly, and given the 1 amp worth of margin of error, you're basically guaranteed to exceed the power carrying capacity of the pins. If you manage to do this across ALL the power carrying pins, you're effectively turning your power connector into a resistive heating element, and this is the result. I was saying this would happen back when the 30 series launched, it was already an issue back then, the power draw just wasn't high enough yet to see these kinds of outcomes. This connector would work IF it was only replacing 2 8 pins, and even then I'd be concerned about the safety profile present. As buildzoid noted, there's a MASSIVE safety margin on the old 8 pin connector, and it's there for a reason as you can see. Normally for anything this critical, you want a safety margin as high as 2 (essentially the connector can handle double what it's rated for), to prevent just this kind of issue. The new connector has a safety factor closer to 1.1, which is fucking nothing. This is irresponsible at best, whoever thought this was a good idea needs to be fired and go back to college, this is covered not even just in EE 101, but we also went over all of this stuff when I took my A+ certification, which is a 1 quarter certification at a community college (or you can study online and just take the cert when you're ready). Literally the most irresponsible thing I've ever seen, and the writing was on the wall the second they announced it on 30 series
We all know an engineer didn't make this decision.
Sounds like Nvidia, Intel, and the PCIE spec group has designed somthing that is GUARANTEED to fail. Cause in Real World application your likely to end up with at least a few connectors not 100% in, with how big the card is and how tight the connector bend would be required to be. So this is a design failure from step 1 it sounds like. And Ill admit Im NOT a engineer, but having THAT tight a margin for error is TOO fucking risky
That large margin is really important once you start thinking about things like insertion/removal cycles, surface oxidation, etc etc. I wouldn't feel comfortable using a connector like that again if it started to melt without being able to see that the metal surface is fully intact and not oxidized; it's extremely common for the same issue to happen again on a connector that already experienced it because it's likely operating at a deficit afterward just due to seeing the high temperature and off-gassing from overheated plastic.
@@big0bad0brad kinda like a good old light switch or power socket after a while. For EU 230 Volts it sometimes is enough to put 10-12 Amps through'em, much bigger and much farer displaced connectors...
I think there is some missing information here. While NVidea spec'd at 9.5 amps the MicroFit line has a standard terminal rated at 8.5 amps and a RMF line rated at 10.5 amps. Interesting from the video was the mention of 40 mating cycles. The 8.5 amp line is rated at 40 mating cycles while the 10.5 amp RMF line is rated at 25 mating cycles. There is an option for lubrication to the terminals to extend the 8.5 amp line to 250 mating cycles. But somewhere the data provided just doesn't match. If they used the 8.5 amp line then thats far less amps then the publication of 9.5 amps and if they used the RMF line then they downplayed the 10.5 amps but the mating cycles doesn't match. All that assuming 18 gauge terminals.
It's most likely the bending of the cables near the connector. In your test, you pre-bent it which is different to what would happen normally.
Realistically, the force is constantly applied onto the cable by the side panel, which may cause the pins to break contact or be misaligned, and it might bend the pins. These factors could cause higher current draw at the misaligned pin, or cause electrical arcs to bridge across the pins, and or depending on the internals, might cause another pin to draw more current.
This would most likely be the culprit I'd reckon.
It’s funny how 1080 ti managed to come up every now and then on almost every channel. That card was such a legend.
got a used one for $450 after 2000 launched... sooooo good
nvidia should stick with GTX
@@DogeRider985 🤣
I'm still running one and have no problem with it
@@michaelingram9354 I have 1 too but upgraded to a 3080 now
Well, I’m sure many seen the Split in the road coming, but thank you for holding the line. Many of the enthusiast, like myself, rely on the right information when it comes to our builds. Keep up the good work and never switch lanes. 🤙
If you're gonna run one of these 12+4 connectors with non-vertical GPUs, you *need* right-angle connectors. I doubt they exist yet, but they need to exist and people need to use them (And right angle connectors even on older cards look great for cable management purposes!)
CableMod are releasing one in a few days.
I personally didn't wanna risk my 4090 or the connector for aesthetics, so I connected the cable in such a way that it has less stress as possible
I replaced my PSU with a new one that has native 12VHPWR connections and cables and there was plenty of room to make the connection to the 4090 without extreme bends when using the new standardized cable. Screw that bulky Nvidia adapter.
Buyers remorse. Not waiting to see what competition offers and 1st adopters experience is exactly what Jay has been preaching.
In the car audio world, the more power and amperage you push through a device such as a subwoofer; you need thicker wires to supply sufficient power current efficiently. This can also apply to a high powered graphics card.
Instead of the 12HPWR connector I would have love to see the Datacenter approach and just see the 12v 8pin EPS (CPU 8Pin) on GPUs, it can Handle 300w and with 2 of them you are at 600w.
And it already is on every PSU, and you can Adapt the GPU ones to it. And it is overall just better.
Yeah, exactly.
Here is a problem with it. Things need to be designed idiotproff. And while the EPS is not a problem in itself the other end would be a problem. (For existing power supplies that is).
Yeah but some cables has Y end with two 8pins on them so it is not fully idiotproof
@@kamilgromek3817 Why tho? Most psus use eps and also the pinout for it.
And Thats just an issue for custom cables, and shipping those is still easier than having 2 different standards.
@@kamilgromek3817 it's even more idiot proof to use the same EPS connector for the board and cards, lol. that being said where the problem lies is changing that standard after the current one has existed for the last 18 years isn't worth the headache.
I guarantee this was caught by numerous engineers throughout testing. Usually this kind of issue is because management overrides their advice.
engineer: “No. No. No. NO! For the love of all that is holy, NO!”
management, upchain: “My engineers say maybe.”
Getting RROD vibes. Engineers told them too.
Quality testers and safety testers: Sir this is a fire hazard we should change it.
Higher ups: Hmmm that'll cost us money and we love money so no,send it like this and let's pray to the divinities above we can blame the clients.
I mean Jay literally showed a screenshot of them (nvidia engineers) reporting the issue
@@SHAdow98V “Will it cost us more to resolve customer complaints, or to prevent the issue?” I dunno, if we start a fire and someone dies that’s really bad. “So it will cost more to prevent the issue, ship it.” 😐
I've been an Aircraft electrician/avionics technician for 10+ years and can safely say that this makes perfect sense. Stress on pin type connectors that compromises pin surface contact can absolutely cause overheating. Scary stuff.
If you really are then you would know that any connector that's not a cannon plug has the same risk.. Do what's so scary?
@@ChuckTownRC51 you must work for nvidia
@@ChuckTownRC51 lol cannon plugs have the same risk. If you really are one then you should agree that Nvidia introducing a product that makes this issue more eminent is a bad thing. The card is too big for that wimpy connector/harness combo.
IMO they would have been better off with two 1/4 inch bolt on terminals and eyelet connectors over 10awg wire to the old pcie connectors......
@@ChuckTownRC51 you can 100% fail to put a cannon plug on properly, typically a large plug with lots of keyways to line up. I saw it happen, caused IFE and immediate impound of a airplane. This was a military jet not civillian. Point is, unless you make sure its seated properly any plug config can do this. Ever see burned pins on a cannon plug, I have, bad seating resulted in heat concentration. Cannon plugs just dont typically melt.
AMD's 7000 series RND3 will NOT have this power connector.
I just got through building a watercooled rig with a 6900xt red devil in the ek block and I kinda like the triple 8 pin, I actually grabbed some 24 pin motherboard cable combs. It honestly looks pretty nice with a good cable run vs this weird 12 pin solution. If the 7000 series is using this plug I'm actually way more glad I built now so all the early adopters can beta test it
I’ve always thought sleeved multi 8-pin connectors coming out of GPU’s looks awesome
If this connector comes on RDNA3 GPUs and if I buy one I might just straight up desolder it and hook up my own solution that ive been thinking about:
Dual XT60 or XT90 connectors as a superior alternative to PCIe 12V and especially to this new joke of a connector that likes to just melt if you "bend the cable wrong".
Would be risky but I could do it
I bought the ASRock OC Formula RT6900XT it has the triple 8's on it as well and it looks sick with it mounted vertically with everything under a custom loop....
@@alouisschafer7212 I’m holding my breath on RDNA3.
Got my 3090 a week ago. Have no problems with the 8pin itself but pls can they not make the connector point down instead of at the sidepanel? Or at an angle like they have done before. I feel like there are rather easy solutions that would solve the whole bend thing. But noone has looked into it because we have a standard. Also a new standard should solve problems of the current standard. Not introduce new problems that are worse.
@JayzTwoCents The only solution I see is one we used in Car Stereo High Amp bends...a 90 degree block that relieves the strain and positions the cable more favorably. you can even connect 2 90s together and have a 180 so your wires are hidden. bending cables at the point of termination is always bad. Good video!
Yes and no: every single connection adds some impedance. Therefore the more you do, the more you add. Why every normal electrician will never say it's OK to use 15 prolongation power filters in a row even if you only connect to the last one and within the specced current?
@@andyshtroymish4997 yes just like 90 degree turns of cable add resistance to electrical flow. This is true in electrical theory, not as noticeable in these applications. Car stereo Amps pull so much power that given the quality of wire and size, it is not an issue. What is though is having a 00gauge wire pull out of a distro block and ground out lol. The same kind of 90 degree blocks used on the 24 pin and EPS as well as 8pin PCIe connectors would suffice. Lesser of two evils
I truly learned quite a bit today from this video. There are many things covered that I had taken for granted. I never knew about contact points and how important it is that there is full coverage. I'm thankful you made this video and doubled down on it because I was one of the idiots who scuffed at your previous one. Thank you, Jay.
I'm getting really tired of big companies that think they can just brush off safety concerns.
And REALLY happy that RUclipsrs like you and Gamers Nexus are calling them out WHEN they do fk up like that.
I love how you bring stuff like this out for us. You are doing the best you can. Love your videos
I laughed out loud at the Intel comment about them helping design the new cable and not using it themselves to burn up the competition
Same here dude. Same time and laughed loudly.
Intel be like: who's running hot now, bitch?!
@@Technatorium Also laughed loudly
That's something Intel would totally do, design a crap connector and not use it, not the most ethical company
Honestly I hope they revise it and recall all cards with it. I like Jay's idea of a bigger connector and they could still work the sense pins into it
Zero chance of happening, Nvidia will just claim it is user error and accept any lawsuit which comes from it, as it will probably be cheaper than recalling cards.
@@TheFreakEnder It is not about law suits, it is about the next cards at that point.
@@Emetsys I want to agree, but they've done as bad things in the past, if not worse, and people are still buying.
That would be highly impractical. But they COULD give owners a rigid 90 degree adaptor (and/or a new cable to go with it) that doesn't have a clearance issue so it doesn't need bent so far.
@@Lord_zeel Yeah they could do that. And now that I think about it, the issue should disappear in a few year as the issue is most likely related to the use of an adapter with 3 cables joining getting it quite stiff and would work fine with a regular cable.
ok I'll mention this first I work for cable assemblies company, we build every thing to IPC class 3 for those of you who understand. looking at what i could find of the amphenol spec for the connector the spec of the wire gauge and the connectors looks ok. but if you bend the wire at the back of the crimp with out any strain relive that is what happens (on any connector) it's just at such a high power rating your tolerance for the damage that can be caused doing this is a lot smaller. they should not expect the end user to treat this cable any differently than any other cable in their systems really and current atx case sizes really don't help for the bend radius needed for this cable.
If Nvidia wasn't being cheap, they could have made the cable longer and even make the connector a right-angle connector to avoid this.
It's not the cable length, but the bend required right at the plug due to case side panel distances. Agreed on a right angle connector, and sounds like Cable a Mod is working on one.
I'm not sure why right angle GPU connectors aren't the norm. Oh wait, right, money. Fucking companies. Who the fuck doesn't bend their GPU cable?
the problem isn't the adapter, it's the card. the card shouldn't need a right angle adapter, it should be built in such a way that the cable can plug in and easily accommodate 35 mm of space without bending afterwards, ideally naturally extending 35 mm without bending. that could easily be done. a connector near the front of the card, perpendicular to the board and facing backwards would allow the power cable to run parallel to the card as it came out. clips that held the cable 4cm from the connector would be sure a good fit.
@@edwardallenthree Nah, you don't design the card around the adapter its the other way around. All they had to do was design a better adapter. Much simpler, much cheaper.
charge you 2k+ for a card and skimp on the adapter ffs. I ain't buying this shit.
Don't forget that the cables come pre-bent for packaging purposes
The best solution for this would be a rigid connector bracket (a bit like the NVlink bridge), to route the connector underneath the card. This way you could plug it in straight from the back without any bending of the cable. You could even put RGB on it to make it look cool
The best solution would have been a 12+4 pin, in the same style as the 6+2 pin GPU connectors or a separate 4 pin much like an 4pin EPS cable, that way the cables retain backwards compatibility (don't need the sensor pins, don't plug them in) as well as not starting fires, it would also help with the adoption of the new PSU's since people wouldn't need to look for specific compatibility, the PSU manufacturers could just switch to the new ones and everyone would be fine and there wouldn't be double the SKU's to account for
Get card burned, create group lawsuit, this connector will never ever will be manufactured again. Also not so bad solution
@@MrTrilbe sure, my solution was more aimed at somewhat fixing the mess which it already made. This connector just sucks and I really hope it gets dropped after this disaster
Or a plug that fits the card to make a safe 90 degree turn since the card is designed NOT to allow the card to be fitted directly in the board and yet still have clearance. For those who mount the card on a riser, the plug can be removed. The design spec should be that the cable weight doesn't cause problems to the connector bend.
The absolute best solution do not get 40 series or any series using this connection because it’s dangerous Period.
The heat generated in wires or connectors is current (Amps) squared multiplied by resistance (Ohms). P(loss)=I²R. 600W going through the connector instead of 300W means 4x the temperature rise in an area close to a heatsink that could be above the ambient spec of the connector. It's not a problem in "ideal" conditions until practical realities are applied. Would be better to plan for more common worst case scenarios. AMD may have already decided to delay using the 12pin connector.
Fantastic. Can't wait for the thin cable adapters from Amazon. Fire hazard.
There's nothing better than cheap under gaged, CCA wire adapters from Amazon!
@@cup_and_cone i like my cheap fire hazard cables directly from nvidia
I bought some Xbox360 power adapter for a weird project I had to reuse an old Power Supply so I cut the adapter to solder my own instead of cutting the Xbox cable and GOD!!! that was weak, looked like 20 awg wires to push north of 200 watts, was affraid to use it. The new 12 pin does trouble me as much... even more if used with any type of extension cables...
@@alexandreapril5694 how did you even compare "Xbox360 power adapter" to a legit power supply..
@@0Blueaura it was to power LED lights, not a computer. For your information I had the same strange feeling when connecting my Cablemods direct fit cables instead of the original Corsair on my RM850X...
They should use the heatsink itself as strain relief and that 35mm buffer. Have the connector point strait up from the PCB and position it so you can depress the clip at the edge of the card and it comes out the top of the cooler.
I can't imagine paying $1600 only for it to fail catastrophically, now he's got to try and get it warrantied and hope the next one he gets doesn't fail too.
No canted cables in future builds 😂 people will be building structural skeletons to keep cable straight lol
he said multiple times that the card works
@@Dreyden12274 Would you trust it not cause a fire? I wouldn’t. So yeah maybe the card works but is it usable and trustworthy? @Sizukun
@@Dreyden12274 Doesn't matter if the card works. With that degree of damage there's no way I would want to continue to use that card if I could get it replaced under warranty. IF being the operative word as I wouldn't be surprised if the GPU company comes back and says it isn't covered as the issue was user created.
@@Dreyden12274 Ya but we don't know if it'll continue to work or catch fire because some of those pins look like they have some plastic melted onto them potentially reducing contact and causing this problem again
Make a followup to this now that Steve on GN has done some real testing and analysis rather than just throwing out theories.
Igor and this dude wil never do that because they just care for rush clicks. Not because they want to find out about the truth.
The pins look the same size as normal board headers, those are usually rated for around 1 amp. 600W / 12 / 6 (for 6 12v connections) is 8.333A. I wonder what kind of testing they did, I hope it wasn't "Trust when the connector manufacturer told us 'don't worry about it, it will be fine!'"
standard current rating is 8 amps continuous per millimeter squared (13 peak) This connectors are so far past safe it's ridiculous
Waiting for Jay's next build with built in fire suppression. It should make R2D2 sounds as it puts out any fire.
He'll be taking water cooling to a whole other level!
Sounds sort of fun.
Trouble is you do not wanna share a room with something putting out a computer fire. Not because there is like water and risk of shock, but for a fire suppression to have any chance of fitting in a computer build it will have to be a Synthetic Gas Fire Suppression. That is a system that stores a liquid, which it turns into gas and puts out fire by replacing the oxygen effectively chocking the fire and hopefully saving the gear. Trouble is of course us humans also like oxygen.
Someone needs to come out with a rigid 90 connector that allows for the cables to re routed in a non stressed manner.
Hard to believe this isn't already a thing. Seems like a very simple solution and product to make.
As someone old enough to have used the 4 pin ketchup and mustard cable Molex connector typically used on older mechanical hard drives I can testify that the pins inside of thsse things move around so much that you can't even plug it in without fidgeting with the cables leading into it. There's also the possibility that some sketchy Chinese supplier didn't supply what they claimed they were supplying, whether that be metals/plastics or the physical size.
@@Mr.Morden it's sad Nvidia didn't just make their connector like cable mods. And instead are beta testing on their top of the line consumers
This is the solution in my head. Nvidia should just supply every owner with various 9 degree connectors. Maybe 1 going down and 1 to the right so they're not torqued. Call it a day.
Whelp statement answered. They come out Oct 31st. The AIB partners should be sending these out for the price they charge. Hell stick the ROG logo on it and you should get 40 bucks
Jay, you said that only Intel does not use that new connector, implying that AMD does use it. However, AMD stated that they are also not going to use that new 12VHPWR (what a name...) connector on the RDNA 3 cards.
just call it 12v high power...what you typed is effectively an acronym....which is still kinda weird anyways lol
If this does start causing actual fires, there could be massive lawsuits potentially involving nearly every major player in the industry -NVidia, AMD, Intel (as a developer of the connector,) board partners, and PSU manufacturers could very easily find themselves in a legal and PR nightmare
Kudos to EVGA for running away from this cluster as quickly as they could.
@@andrewshaw9840 They're in the PSU market still so they can still be effected
Going to have to start wiring up GPUs like Car Audio Amplifiers. Individual thicker gauge wire runs, with individual connections. Car Audio amps used to use lots of plastic plug in connectors back in the 80s and early 90s and as power output started dramatically increasing, they had to go with a safer solution, with individual screw down wire connections.
This would look good IMHO.
i don't see why they have to use such tiny pins for such a chunky power draw, they are insane :D
Frankly the smaller connectors make the product seem a little pathetic. It's like drip-feeding a pot of soup with water. My card has 3 full 8-pin connectors and it looks like a turbocharged faucet because of it.
if it wants 600 watts, it might as well plug into the mains, and regulate its own power on the card, i wouldn't be too fussed running 2 power leads to the back of my pc
Apparently CableMod has an answer -- a 90 degree adapter. It does add 23.2mm to the overall clearance needed to install the card, so it may still be an issue in some cases. It is labeled as the "CableMod 12VHPWR Right Angle Adapter" on their website. Could you check into this and do some testing to see if it may be the answer to the issues you mentioned?
Is it free? Because I just bought a $1600 GPU.
@@SteveL11 Ask NVIDIA to reimburse you.
I was staggered/horrified when I went from a high end 1080Ti on two BIG 8-pin connectors, over to a 3090 on that one tiny little plug - exactly how all those amps were going to squeeze into those little pins. I guess this has been coming down the tracks towards us for a while now the GPU power requirements are shooting up again.
The 4090 takes a good few hundred more watts
It is pretty weird. Meanwhile the 6900 XT uses like three 8-pins, I was looking at it earlier.
This reminds me of what happens when you use SATA-to-PCIE adapters to power a graphics card as they're not rated for adequate current to conform to the PCIE spec. When I fried my adapter doing exactly that (I was in a pinch, took shortcuts I shouldn't have -- lesson learned), it was seated correctly, purely a matter of mismatched tolerances. I'd imagine a similar thing could happen if your PSU doesn't support the massive power spikes these newer cards are prone to, as well. The bending restriction is honestly inexcusable, it didn't have to be this way, probably just considered cheaper to manufacture + ship with a user warning, but like you said, most cases won't allow for proper conformance, so this happens...
Thanks for being a voice of reason. This issue is way beyond you and me - that there should be more of a heads up - this goes way beyond safety. I will personally not use any GPU that has this connector. I am very happy with the performance I am getting from my Power Color Red Devil Radeon RX 6900xt. I choose to use 3 separate cables going between my PSU and the GPU. I have never liked using those "Pigtails" on then end of the cables.
Once I bought a 3080 I saw it had normal PCI e 8 pins instead of the "old" that resembled the new connectors I was relieved, I guess I'm also not gonna buy any of those cards unless I know it's not gonna catch fire
great, now we gonna see 2x12 pins connectors. Every day that passes i got more surprised of how smart EVGA is. Imagine developing something what uses the new and unusual USB type X connector, *and use 2 of them* to work...
One instance in the consumer wild could be a fluke, but this is important to be verified. One instance of a burn or fire means it will happen again at least once. Thank you for sticking to your guns with this. I wasn't upgrading to begin with, but now I'm particularly glad I'm waiting/skipping this gen of GPUs.
There's already a second one
rtx 4090 caliente
I'm perfectly happy with my 6800XT card. Is it the bleeding edge? So far behind it I'd need the Hubble Telescope to even get a glimpse of it, but it does everything I need and it doesn't make my wallet scream every time the electric bill comes to be paid.
Watch the video fully before posting this non-sense. Jay said this is not the only instance. It just went viral.
I was considering a 7950xt but if it comes with this garbage connector I will just wait until rDNA 4. Smh
600W is actually thermal limitation regarding the current flowing throw at certain temperature. Probably @ 30°C or 86°F. If taking into account the connector placement on that card - inside the heatsink, then the ambient temperature around the connector could reach upto 60-70°C. When taking this into account then the max power is degraded somewhere around 400W. This means that @ full power the connector is running on it's very limit and running there in extended time, all sort of things start to happen. So if you want to OC the card a bit or have a bad airflow case it is needed to have separate cooling for the connector (need to add temp sensor to the connector to ramp up the fans). The issue might not be that huge for 3rd party cards if they select the placement of the connector more wisely.
I agree with all that you have said. This connection is unsafe on so many levels... Thank you Jay for doing this type of videos.
I honestly like the look of the symmetry that 3x 8pin and the 24pin has. but i also like that they are trying new stuff. but they should have waited for the ATX3.0 PSUs to give more time for testing
Considering the fact that the 4090 sold out in 2 days everywhere, I suspect there will be plenty of these reports of fire very soon.....this is going to be rough....
After watching Aris's testing I honestly don't think it'll be a big deal. But as both him and Jay said its important to push the connector in properly. Can't really be gentle with it and risk not having a solid connection.
Lots of rtx 4090 still in stock in Norway
@@Sebbz Norway bots must run slower? 😀
@@fitybux4664 no, crypto mining proof of work is dead. So not many scalpers these days
@@Sebbz you must have missed the 4090 launch.... $4,500 4090s on eBay...
8A is a heck of a lot for pins that got shrunk compared to normal PC power pins 😕
(For comparison, the much smaller pins in 0.1" headers are normally said to be good for 1A.)
I like that you and GN and others are brought attention to this, and are going to keep talking about it. If bending the cable like most of us do can lead to this, that is a problem.
I am now rerouting the cabling on my current build. Its not a 4090, its merely a 3070Ti, but the cabling principles Jay has exposed here make me prioritise "correct and safe" routing over "pretty and tidy". Time to rethink some basic safety issues moving forwards.
@@andrewshaw9840 The previous connectors are fine. Nothing to worry about how to route them.
that type of issue is caused by a not correctly seated connector, all the current is flowing through a smaller area rather than the complete contact surface causing it to heat up dramatically at the point of contact. i've seen a similar issue on a car once, a blade fuse wiggled loose but still made enough contact for the lights that it was the fuse for to work, as not all the blade was making contact the connector the points of contact became incredibly hot burning the plastic of the fuse box it was in, replaced the fuse and made sure it was seated correctly and no more issues, customer didn't want to replace the fuse box for a relatively small area of damage (can't blame them it was an expensive part which would require a fair bit of labour to replace) so we cleaned it up as best we could. think the filament of a traditional light bulb, it gets bright due to it being so thin, if the filament was thicker it would not get as bright or would require more juice to get bright, now wrap the filament in plastic and expose to air, hey presto, fire. how the contact on the GPU became an issue is anyone's guess though.
There are more cards/cables with melted connectors popping up. So it doesn't seem to be an isolated defect, but a design flaw.
Sadly, to say I don't see them anytime soon admitting that they have to do a product recall, because of this defect Keep up great work Jay.
And because of that, I don't see myself buying any of their new cards anytime soon either lol. Hopefully other people will choose to do the same.
right angle adapter is a must it seems to avoid bending and to fit the side panel.
That first email feels like a "don't worry, more power through the same connector can't be an issue" and that's super concerning to me. I'm really happy I can avoid using it for a while with my EVGA 3080 ti. This connector really needs a couple more iterations before I'd feel comfortable with it. They should have just made a 16 pin like you said.
It's incredible that card still works with the input slot being so burnt. That would be an immediate rma for me no question
I have no doubt that you have to fight tooth and nail through an rma to prove that it was not your fault since it's not really been noticed as an issue aside from the tech RUclipsrs
its not incredible at all, all thats happened here is theres been a slight bit of resistance, which turned the pins into a heating element basically, its only heated to the melting point of the plastic the copper pins are fine and well under there oxidising point i would imagine. Should be a RMA though since its the adapter that came with the card.
Yeah I have a feeling I'll be skipping the 40 series GPU's completely. If price and having to upgrade my PSU weren't already major barriers, Nvidia's refusal to admit this is a huge safety issue seals the deal. Thanks Jay!
Now I'm actually glad I bought 30 series card.... I'm also going to skip 40 series. They should build better connectors, these tiny connectors, were always looking too flimsy.
@@Slav4o911 Same. Got a hell of a deal on a 3080 12gb so I feel like I'm set to skip at least this generation.
Probably no reason to if someone has 3080/Ti or the 6950XT types.
I imagine before long they will be forced into a recall if there is fire risk and see enough people having it happen.
All it's gonna take is someone's house burning down. Just a computer burning down is something Nvidia could write off and keep hush-hush.
Come to think of it, scratch that, it would have to be some MILLIONAIRE'S house burning down.
Because we all know poor people can't afford good lawyers especially after their house burns down.
And good luck keeping some entitled millionaire quiet with "hush money" unless you pay them like 10x more than their house (and all its possessions) was worth.
Me? I'd start a fucking class action lawsuit and the terms for settlement would be as follows:
Nvida buys back all 40xx series cards that have the 12vHPWR connector on them, no matter if it's a founder's edition or which AIB made them, and then re-designs all the cards (at its own expense, including the cards made by the AIBs) to support the previous standard of 8 pin PCI-E power connectors in sufficient quantity to handle the expected power demands, because like Jay said this is and was a (bad) solution looking for a problem.
In other words, 'Nvidia, I hope you know you're going to pay for this' meant in the literal sense, since it seems that's all the CEO understands.
all it will take for one person to have there house burn down and sue all parties
@@tacticalcenter8658 Things that cause real fires don't tend to let themselves just get swept under the rug like all this other stuff does. Fires kinda burn themselves into people's memories.
If the 5000 series cards aren't significantly cooler (and by that I mean more efficient) than the 4000 series cards, then I'm gonna be looking seriously at AMD's offerings for the first time in a long long time. Last time I bought an AMD (then still ATI) graphics card, AGP was the accepted industry standard graphics card port. So yeah it's been a while.
Oh yeah and apparently this year marks the debut of the "space heater" feature on a gaming PC. All you have to do to enable it is to have a 1600w psu, an am5 motherboard that has a 7950x in it, a 4090, and somehow have the 12VHPWR connector magically not melt.
For cooling I'd suggest a custom water cooling loop with 4x 480mm radiators to handle all that heat. As for a case, what, you want a case? Good luck with that buddy, go get yourself a test bench type case instead and just build a roll cage around it with angle iron and a mig welder so that you don't fall on it and break all your expensive stuff.
@@tacticalcenter8658 Dude, I was making a joke when I started talking about "space heater PC", and it was meant as a "normal space heater" not "fire hazard space heater" kind of joke.
As for the 12VHPWR connector, if they keep burning up a recall's gonna happen, I guarantee it. If the recall doesn't happen, a lawsuit will happen that will force the recall.
And with that recall, will come a real re-evaluation of the design constraints around the 12VHPWR connector.
In any case, I'd prefer an XT150 or Andersen Power Pole connector over that connector, who needs all these itty bitty pins when you could have 2 much larger pins that are a lot harder to cause to have a contact problem, or at least that's how my thinking is going. Heck, give me studs to mount ring terminals to, anything other than this connector that is ridiculously small for the amount of power going thru it.
Plus, that whole "35mm no bends" thing with the 12VHPWR standard is an asinine design constraint, if that's the constraint you have to throw out, maybe don't mount the fucking GPU's power connector to the "top" of the card, ya idiots! That way we can use the fucking cases we already have, instead of having to re-invent literally everything (looking at you, 4090 support bracket) just to cram in some "obviously someone cranked the scale slider too big" video card.
EDIT: And yeah, there are several things wrong with me, but if there's nothing wrong with you these days then there's something wrong with you because there's nothing wrong with you, because this world's so fucked up that everyone's getting messed up with it in some way like it or not, it's unavoidable.
Yeah I'm calling it now. Nvidia will end up doing a recall at least for the 4x 8 pin (or 3x 8 pin) to 16 pin adapter. I have a 4090 and will be building my new PC later this week but I luckily bought a new ATX 3.0 PSU so I don't have to use the adapter at all.
Only half the issue, even if you don't use the adaptor you still can't bend the cable. If you do, you have the same issue
also with no adapter, you have two connectors with very small pins and potential heat up ...
How's everything going after 10 months? I'm using the cooler master psu wire but currently not near home to verify my temps
@@saadhero9107 I haven't had a single issue so far with my PC I built last October. I just pre-bent the cable a bit and let a little length of the cable inside the main part of the case.
Looking at the difference between the 12vHP and the older PCIE 8-pins is the depth of the seating.
The 12VHP seems to have the same depth as the 8pin, which certainly contributes to the issue here. If they beefed it up a little more, we could probably end up with less of these issues.
Sad it's even happening period. Shows Nvidia's true colors.
I am thinking the connector was made for rackmount servers were the cards would always be on the side but on their side is more akin to saying they will be 1/2 way into the case that is as long as a full ATX case is tall. So the issues isn't there. nVidia has the issue as they put it onto consumer cards not just cards meant for the server market. Consumers are well doing what consumers do and running stuff in ways the engineers never thought it would be run. Now we have failures just hopefully not expensive failures.
Current issues often still only burn the positive pins, at least when it comes to GPUs and motherboards but this might be because the ground is usually all tied together, slot and cable, and there's A Lot of grounds on the slot. the positive sources are often isolated, with some VRMs (for memory and the like) on the slot only and some on cable only.
Thanks for keeping a lot of us informed Jayz, you're doing a good job. Your assumptions and questioning are absolutely right and justified. The high current and low pins contact surface problem you're talking about is totally accurate; I see those kind of problems in my job everyday (I'm an automation technician + master electrician). Stop feeling like you need to justify yourself to those assh... They exactly know this problem for sure by now and they are just pissed at you because you're showing it to daylight. Those guys are so afraid of the amount of money it would cost them to recall all those 4090 that they're trying to bury it under a sh...tload of BS. And the worst part is they don't give a flying f... if people PC and even house catch fire. All this due to a bad design; everybody with a brain knows that power cables are bends in every pc that exist, especially with this size of a card.
I'm reading that AMD 7000 series cards will not be using this connector. Article up on Tom'shardware says they will be using the "tried and true 8 pin PEG connectors"