NVIDIA Responds to Melting Cables, Warranty Concerns, & 12VHPWR Adapter Failures
ฝัง
- เผยแพร่เมื่อ 4 ก.ค. 2024
- Sponsor: Fractal Pop Air RGB Case on Amazon geni.us/TMxbP
NVIDIA has responded to the RTX 4090 12-pin power cable concerns pertaining to melting / burned 12VHPWR cables. The company has provided a statement on the situation, given a failure quantity, and has commented that it will be replacing all affected units under warranty. GN was able to source new information from board partners pertaining to NVIDIA sales count and, using this new 50-unit number provided by NVIDIA, we can now get an accurate range for failure rate. We also talk about how to properly install an NVIDIA RTX 4090 or RTX 4080 and how to properly connect its power cable. The tutorial at the end shows correct 12VHPWR (12-pin, "16-pin") cable installation for these GPUs.
Watch our NVIDIA RTX 4090 power cable & adapter cable (12VHPWR 12-pin) investigation here - part 2, conclusions: • The Truth About NVIDIA...
Part 1, attempts to recreate the problem: • Testing Burning NVIDIA...
The best way to support our work is through our store: store.gamersnexus.net/
Like our content? Please consider becoming our Patron to support us: / gamersnexus
TIMESTAMPS
00:00 - NVIDIA Responds to 12VHPWR Melting Issues
04:17 - How NVIDIA Determined the Cause of Failure
06:30 - NVIDIA's Warranty & Replacing All Burned Devices
08:42 - On Cable Validation Suggestions
10:09 - Foreign Object Debris
11:11 - How to Improve the 12VHPWR Cable
13:04 - How to Connect RTX 4090 & 4080 Power Cables Correctly
17:10 - Last Piece of Advice on Cable Management
17:40 - Conclusion
** Please like, comment, and subscribe for more! **
Links to Amazon and Newegg are typically monetized on our channel (affiliate links) and may return a commission of sales to us from the retailer. This is unrelated to the product manufacturer. Any advertisements or sponsorships are disclosed within the video ("this video is brought to you by") and above the fold in the description. We do not ever produce paid content or "sponsored content" (meaning that the content is our idea and is not funded externally aside from whatever ad placement is in the beginning) and we do not ever charge manufacturers for coverage.
Follow us in these locations for more gaming and hardware updates:
t: / gamersnexus
f: / gamersnexus
w: www.gamersnexus.net/
Host, Writing: Steve Burke
Video Editing, Trainer: Mike Gaglione
Trainee: Video Editing - เกม
Learn about the data in our testing video! th-cam.com/video/ig2px7ofKhQ/w-d-xo.html
Use code "THISISFINE" on store.gamersnexus.net for 10% off til next week!
First a small nzxt h1 v2 case then nvidia 4090. Prove that they have never tested their products
11:33 Make end that is inserted into the socket in bright colour for eg. red. When you don't see red, it is fully inserted.
@@njgrant3988 and Gigabyte IED PSU.
I'm worried that there will be more "This Is Fine" moments occuring in the future. Probably a burning MB, gasoline filled liquid cooler or thermal paste made from mercury/lead.
Perchance is the resistance that is felt more due to the friction of the plastic on plastic or is it more due to the metal in metal. If the former, could they also 'shave' the male keys smaller?
@@First-Name_Last-Name my money will be safe until there is another good product i can buy
Nvidia didn't respond this whole time because they were waiting for Steve to finish the investigation and do the work for them.
you’re not wrong 😂
Yep they just outsourced the testing to GN without notifying them
Like when game publishers push out a game and wait for modders to fix it.
Steve: Releases deep dive video.
Nvidia: Write that down, write that down!
that or more plausible reason is that customers prob woudn't want to hear that they (users) are at fault from nvidia.
Imagine the public opinion of Nvidia blaming user error if GN hadn't released their video and research on it already
Almost makes you wonder how and why the information happened to come in that order within a couple of days, after weeks of speculation
right the optics overall are just much better this way .2nd source with a customer first mindset. rather than a corporate first.
@@rillepp Why it took multiple labs teh same amount of time to gather evidence and make an analysis?
Steve Jobs did it....they gave a rubber band their fanboys and everyone was well....Iphone4. PPL STILL kept buying it. Nvidia doesn´t care.
If GN didn't release a video and NVIDIA stated their conclusion, the internet continue bashing on nvidia in their little echo chamber.
I'm convinced GN has had a big impact on tech companies becoming more consumer friendly.
I think he does but I think he should’ve been even more harsh and aggressive towards nvidia so that they give us free stuff cuz nvidia owes us
well it sure isn't looking that way
@@psychoticgiraffe What do you mean by "free stuff"? Nvidia won't just give everyone a bunch of cards because Gamers Nexus criticized them more.
But big tech corps are the opposite of "consumer friendly"..
@@psychoticgiraffe Nvidia does not owe you anything. The world in general does not owe you anything. That you are owed something is a weird parasocial assumption.
As somebody who's been working with cables and connectors for about half a century now, my usual advice is pretty much exactly what was said in this video, but actually with an extra step at the end:
1. Plug it in as firmly as you can, preferably until you hear or feel a "click"
2. Pull it back out with the same force you used to push it in. If it comes back out, go back to step 1 and use more force.
3. Do all your moving around of the device, cable management, etc.
4. When everything's where you want it and you're ready to close up, go back and push the plug back in again as firmly as you can.
5. Pull it back out with the same force, and make sure it won't come out.
6. Push it back in again one more time.
7. Close everything up.
I still say that the only reason this sort of thing is a problem in the first place is because the connectors are actually badly designed and are not actually really capable of reliably carrying the power that they're trying to put through them (this application vastly exceeds the specs of even Molex, who designed the original connectors). They do _technically_ work, for the most part, but only because they've pushed everything so far that there are just no safety margins left, so if any small detail is not absolutely perfectly correct, you will get problems. That is just not how a well-designed connector should be, particularly for something dealing with these massive amounts of current. I'm still actually really concerned that even when people have them perfectly seated, we will start to see more and more problems in the future as these connectors age and the pin connections become slightly less perfect over time just due to normal use. Only time will tell with that sort of thing, however.
Don't be too shocked if a year or two from now we have another round of all this with people who actually did everything correctly and thought everything was fine until something suddenly melted down with no warning, though...
(To be clear, I don't blame Nvidia, at least not entirely. I mostly blame the PCI-SIG for this whole debacle.)
The whole pci connector needs looking at. I can't stand pigtails. 2 wires into 1 crimp terminal is the worse.
👍
Agreed. PCs have crept into a power envelope where internal power transmission isnt necessarily adequate anymore and now we begin to see the fallout from that.
Thanks for the write up. It's kinda weird how many comments I'm seeing of people just making fun of random strawman users, when in reality the connector could really use a revision. The tech community has gotten very weird these past few years.
Even GN staff in the last video was making fun of strawman users, something about "typical Fortnite users"
the pci sig has this connector added to atx 3.0 standard beginning of 2022
but remember that nvidia actually made the 12 pin connector for the 30 series launch
i remember them saying how proud they were for their new connector
so how i see it nvidia is to blame their design is sub optimal and now it was even added as a standard yikes
The insight and wisdom around 9:00 is exactly why I consider this channel a gold standard for tech journalism
Thank you for the kind words!
It's actually a Titanium standard.
@@thingsiplay Surely, you mean "Prime Ultra Titanium"? :P
This and Hardware Unboxed.
@@jemborg yeah, if it wasn't for HWuB ridiculous bias against anything made by Samsung.
The Nvidia statement really corroborates GN's in-house testing and proves them completely correct. Great to see how reliable and trustworthy this channel's content is.
Pretty sure Nvidia did zero testing on their own and just told the PCI folks to get on with it. You can thank GN 100% for this. 1 TH-camr found this before a billion company? Really?
@@tqrules01 nvidia doesn’t care… but they make good gpus, so they can get away with it. Sad really…
@@tqrules01 lol 'pretty sure' meaning it suits your feels based on zero research. right? Obviously the opposite of the reality but do you bud, bless your heart
Or NV just let Steve do free research LOL
@@kainhall MAX PROFFIT.
Personally, I would move the latches to the sides, so you squeeze the connector to remove it. One latch in the center doesn't make for a stable connection, it can rock too much. Also makes removal easier, squeeze the sides of the connector and pull, less strain on cables.
remember the dvi cables with the screws left and right? why not something similar with much smaller screws that force the connector into position. might seem a bit "excessive", but id prefer it over a system where i can fuck up this bad even if i double check. just because this weak ass plastic clamp does not snap in properly
@@paulpur3 We use screwdrivers for installing components in computer so why not make power connector with actual screw. They can be quite low profile and perfectly secure at the same time.
I mean we are talking about cards that cost north of 500 bucks nowadays having more that just bent metal pins and a plastic sleeve is not asking too much I don't think.
at this point they might as well just have left the PSU plugs on the PCB >.>
A latch system like you have on memory dims would be an idea where it is quite clear if they have engaged or not. I have seen similar designs for automotive applications where it is clear there is a lock
@@Vaasref Anderson connectors. That's the military style you're referring to. The problem is they're bulky, complicated, and expensive. Remember, these connectors aren't just meant for graphics cards. They're an industry specification. So, we could see low end Motherboards using them instead of 8 pins.
Mad respect to your in-depth reporting/analysis. Over the months, you guys have become one of my favorite go-to news channels.
Good to see Nvidia finally communicating clearly and assuring 4090 owners on any warranty concerns. It took them long enough and cudos to GN to arriving at the same correct conclusions earlier than Team Green!
That statement needed a -thanks Steve at the end lol
Glad they're finally making a statement!
To be fair they couldn't communicate because they had no idea what the problem was. They found out when Steve did his previous video.
Well, they probably would have lost a lot of 4090 sales if they hadn't. They still might.
NVIDIA was too busy counting the over $200 million they already made on the 4090 sales ALONE. With sales figures like that there is ZEO incentive to lower the price to realistic levels.
I hope that analysis lab contact is gonna continue with their cooperation on further investigations. Very well done, interesting watch.
Completely agreed! Loved that collaboration.
@@GamersNexus I wonder if Nvidia knows that people like me hate their Adapter Squid from "20,000 Leagues Under the Sea"? Thanks, Steve...🇺🇸 😎👍☕
I never thought I'd see the day where there's an 18 minute video about how to plug in a connector. Amazing.
In safety nothing is trivial. You see this all the time where even the smallest thing can have a huge impact and becomes the weakest link in the chain.
Welcome to modern tech design.
Well with your description it's ridiculous, but obviously the video covers far more than this.
Thanks to the entire GN team for the hard work delivering accurate and trustworthy reporting. Great work.
In order to plug the cable in correctly, you have to push as hard as Gamer Nexus has pushed the frontier of publicized benchmarking and thermal testing. The work with the 4000 series was insanely in-depth. You guys are truly putting to shame all the hardware journalists in the world. I've never seen a magazine or other TH-camr go to the lengths you have. As my 4080 will arrive on Monday, I already thank you for the peace of mind and the hard work it took to get there.
😄👍
Gold comment
Isn't LTT Labs supposed to be the ones doing this level of investigation? They made millions off their backpack alone and have nearly 100 staff in multiple buildings.
@@sammiller6631Theyre too busy talking about 6 foot widescreens nobody will ever use
@@sammiller6631 they bought the test tools.. Maybe not yet fully setup..
I think a simple solution is to have the tip of the connector red, so if you see any red around the socket when the connector is inserted you know it's not in all the way
Or maybe just use one of the thousands of available connector designs that positively latch, i.e where the latching pushes the connector into place? This is not a $19.99 card where they have to shave off every cent of cost. They can afford to use a connector that costs 2 or 3 cents more.
GN it could be seen in GNs videos that the connector was fully in. It just didn't latch on. So the "Red" color would be pointless. And to rely on Sound for the Latch to snap, is also bollocks. Because most PC connectors do not make a sound to begin with.
So all you can do, is put it in with pressy and wiggle the connector around until you can't pull it out anymore. Do i have to mention that this can damage the Plug and the Socket? No matter what and how, this GPU Connector is complete horsecrap and needs to be removed!
And i can already see how people like GN, who constantly switch out GPUs plug and unplug them. Down the road, these connectors will wear out. This was a pretty big foreshadowing in the last GN Video. In my opinion, this is the most likely scenario to happen.
The simple solution is to change the pin design and mount location.
These pins flat out suck
I would prefer a solution where we just bump the PCIe graphics rail to 48V, and stop driving tens of amps down user-installable cables.
it seems like it can be practically inserted all the way, just not latched, I think that this solution wouldn't work well. The sense pins are better because if it starts to work it's way back out of the socket after it's plugged in then it will alert you digitally, where as red paint still requires that a use manually inspects the cable repeatedly in the future.
you're the only channel I've seen do accurate, non-clickbait, non-sensationalized reporting on this issue. Really appreciated!
Thank you, Steve and GN for all of that legwork to make our builds and experiences alot easier and enjoyable. You all rock!
Plug it in, with clamps, put cinder blocks on top, and then epoxy the entire thing
Hot glueing cables suddenly doesn't sound that bad of an idea XD
Or just plastic weld the two together.
Or just plug it in like any other PCI or EPS connector
@@thelegendaryklobb2879 Even better-- Buy cables made out of glue sticks and let the 12 pin connector do the job for you.
All of these findings makes me terrified for 4000 series cards being sent out in pre-builts. I feel like a lot of new or unexperienced gamers are going to be getting hot glue or unlatched fire hazards in the coming years
I for one welcome the return of our screw based friends that we've lost along the way, R.I.P. VGA, serial, and parallel connectors :`-(
@@msytdc1577 get screwed with your screw fetishism
Total design flaw issue with lack of proper education for users about how to tell the cable is fully seated. Hopefully Nivida will be on an education campaign and changing any stickers or attaching one to future adapters.
Yes .04% of them could suffer SERIOUS DANGER of a melting connector!
you’re the first person (that i’ve seen at least) to bring this up. if some OEMs/SIs can screw up the most basic things, then 4090/4080s are gonna be a total nightmare to deal with.
Extremely grateful to have gamers nexus as a resource. Thank you everyone in the team.
I like linus and Jay but Steve and his team make sure they are 100% sure of what they say because they do the work, thank you always guys your work is 80 plus Titanium
I've chosen to watch less the former and completely stopped with the latter.. haha
Steve and GN are much more informative/factual
Linus and Jay seem to only make mainstream videos to generate some controversy instead of technical stuff that add value, and always have a "holier than thou" attitude towards pretty much everyone else, at least from my perspective. I do no see any of their videos anymore.
Other channels that I watch are Hardware Unboxed, Daniel Owen, Paul's Hardware, and if you are interested in seeing how GPUs are dissembled and repaired, check out northwestrepair.
@@RatBotEins I'll check them out thanks😁
In the medical device industry, we don't use the term "user error", rather we use the term "use error". This shifts the blame from the user to the design. It's not uncommon for something like this to slip through, what's important is the response and corrective actions for product that's already out there and for future product. Shortening the sense pins, probably on the card side, sounds like a reasonable solution. Let's see what they come up with.
I am absolutely shocked that a video has to be published about how to plug in the connector right. It means for me that this design is not good and possibly could be examined by regulatory bodies to determine if it can be considered safe enough to be used. Although I assume it has gone through such examination and got approved but it looks like practice proves otherwise. Thus it could/should be re-assessed.
@@sandornyemcsok4168 0.04% failure rate? 50 in 125000 cards? Not safe enough to be used? Okay, sure. I'm certain you're using 50 things with failure rates oscillating around that chance of failure or much, much, much higher %.
@@TheDravic Those numbers make the very incorrect assumption that all of those cards have been plugged in, they haven't, thousands are on ebay and other sales sites being scalped and thousands more are waiting to be scalped. Half of that number at best may have been opened so far.. so your real failure rate is closer to 1 in 1000 which is too high when the failure has to do with something that is supposed to be standardized. Would you wait until a failure causes the death of an entire household before changing from a error prone and objectively badly designed connector?
@@bdhale34 you don't have any evidence to support the numbers you just made up in your head. "Half of the cards have been opened at most". Listen to yourself lmao.
The truth is right in front of you. If you want to make numbers up, at least have them be within a realm of plausibility. Let's say the real number is 10x bigger. That's 500 instead of 50. This would move the percentage from 0.04% to 0.4%. Crazy :)
That would still only be 4 cards for every 1000 cards.
It is user error
Love that you contacted NVIDIA to confirm the RMA for user error, it was literally my first question after reading their statement.
I think the best way to test if it's plugging in is to pick it up by the cable, and sling it around your head like a nun-chuck to see if the card stays in or goes sailing across the room.
"No, no. He's got a point..." is the meme I can feel from this comment albeit very wrongly test wise and dangerously to feel the plug work properly as intended
I think the most important factor about this issue is that it wasn't an issue before this.
nvidia or who for them have done test with normal ppl not dum af guy that don t know how to plug a cable
Isn't this PCIe 5 connector basically FOAK?
IMO GN should have not made any of these videos because nvidia should’ve had to investigate this themselves and they owe Steve millions of dollars for his work; he single handedly saved them from an expensive recall; I personally think the cables should be recalled still, we should be able to get a free cable. Nvidia is scalping us practically with their prices and can’t give us a 50 cent cable it’s ridiculous, instead we get a cable worth 2 cents from China, thanks Jensen
@@psychoticgiraffe Nvidia did their own investigation as well lol
Great work on the last video, can't wait to hear Nvidias response.
NVIDIAS´s response " Thank you Steve. Now go buy our cards."
Its really nice to have a pc hardware investigation channel to learn the facts from.
Like the fact that a car can't run on pure sweet fat of the hog.
Why am I so obsessed with this coverage, I don't even have a card from this decade.
You guys should be very proud of this work.
Amazing reporting.
I think alot of the problem is the new design puts the socket so far into the gpu that you can't really see if you're seated or not.. The old sockets protrude away from the gpu and are much larger so you can actually see if you are seated or not.. This new design makes it hard for people who have poor eyesight like me to tell if it's seated or not
Honestly even with good eyesight it can be hard to tell because it's usually a black cable going into a black plug in a dark computer case.
the new design is so bad that i never had this issue on 20 years of nvidia gpu, thats how bad it is, they owe us free non-fire hazard cables with an L design that is foolproof for this oversight, otherwise someone is going to burn their house down rendering 8k video
@@psychoticgiraffe Or you could just not be dumb. IDK.
Not even that. How much Force you actually need to push it all the way in is even more stupid if you realize that gpu is actually seated on your motherboard.
That's what it clicks
Nice to hear nV finally respond. They finally scoped the potential damage at less than 0.05% and decided, sure we'll cover this issue! Good work, Steve/GN...you damn near finished that plugs cycles in one video for us!
It's incredible just how high the percentage of people who had this failure posted about it online. Personally, I think that's awesome. And I have to say 0.05% is lower than I would have thought. I would have fully expected closer to the 0.5% mark, but that may just be the amount of people who initiated a return. Guess the problem is in fact user error if the number is that low. Most people getting these gpus have plenty of experience and I guess weren't used to the occasionally high force required to plug the connector in compared to the 8pin.
To me this ultimately feels like a design failure. Sort of like the 24pin. Some of those straight up DO NOT GO IN. I have cracked traces on motherboards from the amount of force required to get that thing fully seated. The only method I use now is the slow and steady wiggle back and forth method until its fully seated such that you can't get your fingernail in between the connectors and the latch has been engaged.
Also interesting to see there was more hair on fire than melting gpu's over this when the data comes out and shows how little of an issue it was to begin with. But if you listened to all the chicken littles out there you'd think 4 out of every 3 40 series were melting down.
@@mattk6827 Still not acceptable, over time the connector will become worn and brittle (prone to failure). Connector definitely needs a redesign.
After watching your video on adapter burning up, I checked my 4090 and it was not in all the way and I had originally pushed it in very hard and could have swore it was in all the way-I pushed very very hard and it went in fathther and could feel a click and I have been building computers for 5 years.
Excellent PCA on showing the proper seating Steve. I appreciate the work you and your team do on these type items, without your reporting likely Nvidia would not have been so quick to respond.
I made this comment on the original GN video: if users can think this connector is properly inserted when it is not, that's a design flaw. It should be almost impossible for the connector to be inserted improperly IF such improper insertion can cause melting or fire. Improper insertion should cause the system to throw a VGA error, thereby alerting the user to the problem.
Yup. Funny to see all the people saying you deserve to have your card break if you didn't put in the cable right.
It has literally never been in issue in my lifetime. No connector I've ever used could become a fire hazard if improperly inserted.
@@nathanielletourneau9952 Indeed, and none ever should.
Exactly what I commented on.
A cable should be socketed and done forever, without the need to check on it every time the user removes the side panel or pushes the cables to cable manage.
Every other direction or “instruction” is there to cover a poor design.
And the engineers know perfectly well that even power users do not assemble PCs using checklists and redundant checks.
Maybe color the tip of the connector so you know when it's fully seated. Also makes it easier to see how far it's in since it's not just all black
Or, just maybe, use the latch?
Well plugging in a connector fully hasnt really been a issue till now. and form the marks on the connectors those connectors have been barely plugged in.
50-ish users didn't give it a tug, after plugging it it, to check if latch got engaged. No eyeballs required.
50-ish users didn't even do the easiest thing, and you expect those people to look for some half a mil line gap (of different color) in between the male and female ends, in the final built system.
All that adding complexity to the adaptor (2tone plug) would do, is raise the cost of production and not get included in the box> sold separately for way more; and those 50-ish people would still not double check their work :/ . Even if it's not 2 tone, black is still the cheapest by far. But if's not in the box (expecting it to be included), and _you_ NEED it NOW, _you_ 'd just A Prime the cheapest and fastest anyway :) (probably not the limited edition Official Two tone one, but a lesser AWG knockoff); because _you_ probably are not the guy that checks the SIG spec anyway. _You_ 're the .05% guy that didn't give it a tug.
"Invent a fool-proof design, the world just comes out with a better fool."
PS: If you have a sharpie you can put the line in yourself, to check your connection ¯\_(ツ)_/¯ .05% wouldn't even give it a "tug-tug"
@@patrickbateman7444 yeah. This is kind of PC building 101 honestly. It seems this 0.4% is mostly user error which is not surprising.
@@IcecalGamer Way I see it is if you aren't willing or smart enough to double check the cable is plugged in on a pc part you just spent 2 grand on you deserve to melt your card.
The deep dive was a good video. Very informative. Keep the good work.
Thank you for providing such high quality coverage on this topic!
Big thanks to the GN team for always being the first to go literally beyond normal testing and giving the cause and how to fix it!
It is AMAZING that you were able to hit the mark with this reporting before even NVIDIA could get this info out, and they said exactly what you said. I wonder if your video release had an impact on them?
Incredible, outstanding, top tier work from the GN team!
I, for one, wonder if they haven't just used GN's research instead of their own. I guess we'll never know.
Its different. Its very likely NVIDIA has enough evidence to make the claims but they need to be 100% sure there isnt any other variables. GamersNexus can make a post whenever they make findings but NVIDIA has to be very very careful what statements go out as it can seriously affect their brand.
GN did an incredible job though either way and im glad this rumour nonsense can finally be put to rest, reddit gets unbearable at times lmao.
@@chrisrosenkreuz23 GN pointed out that there are a lot of approvals that need to be in place for this kind of release, and that the investigation wouldn't proceed at that much of a different rate to their own- the similar release timing being indicative of that.
@@MFMArt honestly speaking, people will find other things to bitch about very soon xD
They're already at it with "Nvidia did nothing, user error LIES" and so on and so forth. This connector simply put is designed by people who know what they are doing, they didn't expect people being THAT "stupid".
And seriously tho, the fact that you can RMA your card for something that YOU caused is pretty damn good (to be fair, Nvidia probably don't really care about 50 cards more or less tho so yea).
Humans are difficult and annoying at times. Sure Nvidia is currently not looking too good especially with their highly priced cards and such and you maybe don't like their work, but seriously stop being a bitch about everything they do.
I myself thought that this connector is a bit weird, handling that much energy from so many connectors, but apparently it can very well do it. Just use your cable correctly Karen
@@Mom19 You wouldn't really want to deny any RMA for things like this since you're potentially losing a customer willing to buy your (expensive) product(s).
There're times where just honouring a RMA costs you less than denying a RMA.
It would be interesting to see/measure the amount of actual force required to put these kinda connectors (and others too?) compared to just how much force the boards themselves can take before they suffer damage in some way. It'd be fun to find out we're worrying over nothing when pushing connectors in, but also fun to find out we're really pushing that red-line :D
The best is to look at the clip from the side. The top should be parallel to the side of the plug (i.e., the clip has latched successfully to the plug). If the clip appears to be slanted, then the plug is not fully seated. This is how I check the plugs are inserted fully into a modular power supply and the motherboard.
@6:20 what it was fully inserted, but not latched, it's IMHO more design error than user error.
This is the issue I see happening. Fully inserted but not lached and add in case side panel and cable management like he mentions and boom failure. Nvidia should have really designed the sense pins better to not work when it becomes so loose it can melt
I wont be happy until jensen sends me a leather jacket with the power cables/connectors. HIS LEATHER JACKET
His oven too!
😂😂😂
@@PSYCHOV3N0M and the oven mitts 😂
And his shoelaces!
Sounds like you have set yourself up for eternal unhappiness ;-)
@@PSYCHOV3N0M i take what in the oven pls! :D
Thank you and your team for being a great source of fair and informative testing. GN's efforts have made it a highly valuable consumer protection group as well as an entertainment and instructional producer in the computing world.
Great to see Nvidia validating the great investigative work done by GN. Shows the quality of work and testing being done by the GN team and delivers great viewer confidence going forward. EXCEPTIONAL WORK GN🥰👍💪. Respect ✊✊
connector realistically needs to ensure that the signal good is not triggered until: 1) fully mated contact engagement AND 2) latching mechanism is secured. It is possible for 1 to happen while 2 not happening could result is loosening over time/vibrations.
wow. i got a 4080, and i know its not as likely to burn, i had issues getting it to boot at first because my 12+4 wasn't plugged in all the way. i had it plugged in maybe like half way, but that felt normal because it was really tough to push it any further, so i assumed that it was normal. i had even unplugged and replugged it in during my troubleshooting, and still couldn't get it to work. eventually i did realize that it had to be pushed really hard all the way in and it started working immediately. i think even having a picture of how it should look on the packaging would make this way less of an issue.
Man, this is some seriously impressive and very important investigative journalism. Thanks Steve!!!
Thanks Steve and all involved at GN for these excellent video's! It's gr8 that Nividia are going to do something about it and I can conclude from the video and that what was said about users not making sure the socket and plug are firmly seated and fully home. I agree with Arudanel below about the connection being weak! I did hear the click of the plug in the video but this doesn't mean its fully home and it was still looking flimsy with a bit of play this may lead(no pun intended) to the connection shorting out! the socket still moves slightly so my observation is that users cant see that its fully home and so maybe a cutaway section that enables the user to see the connection and also a much better secure fitting than that that is available like a flange if the flange is flush with its co flange a green indicator led lets the user know the cable is fully seated if at any point on the flange is not connected the green led wont come on if it don't come on you cant use the device as a fail safe mode. also a temp sensor to test for any heat outside normal working conditions. Failing that the other way is to have a cable connection from device instead of the buried on board fitting on the component board a bit like an extension cable fitting this also would for sure be a visible sign whether the connection is fully home with a secure fitting to make sure it is done correctly and safe and As mentioned earlier if it is not, there will be a light indicator or a warning audio signal. Only when it is safe will the device work and not before! you should always check fittings in case they have become dislodged during a build and definitely before booting up the system. A word of caution here there are fake parts so only get from certified authorized retailers and if in doubt leave it out! and contact the manufacturer for a part. take into account even the manufacturers can be duped as so we all so again it's down to everyone to verify everything is what it claims to be! This may take time but not checking has its consequences stay safe everyone and happy computing/gaming and have a MERRY CHRISTMAS/THANKS GIVING or HAPPY HOLIDAY AND A VERY HAPPY NEW YEAR 2023!!!😁🤗😘😘🥳🎶🎵💃🕺🙏😇
I'm curious if the 12VHPWR connector has any potential issues on the PSU side with new PSUs that have the connector? It would be much harder to see if the connector is all the way in on your PSU than it is with your GPU.
Yes that is possible, although there aren’t as many people running the new power supply standard with this connector so the sample size is much smaller.
Pushing isn't just for swings and shopping carts, and be sure to push deep and push hard ;-P
PSUs manufacturers tend to make their own female and male conectors to their own propietary connectors, like corsairs Type 4 cables and so on.
@Gamers Nexus I advice you be wary of Users commenting like User here: msytdc ,which comments are offending and demeaning also tending towards personal insult ,especially to an esteemed and known Content Creator ,as is TastyPC .. Please Resolve that issue.. if you be so kind!
@@yorchli3152 That's true but more and more PSUs are being released with a 12VHPWR connector on the actual PSU.
i started my IT journey this summer. I am taking my A+ soon, and will then start the rest of my certifications and coding classes. Watching you theorize, test, diagnose and fix issues is awesome. you obviously have a wealth of knowledge and comprehension, but you make it understandable and easy to digest for outsiders and amateurs. It's really inspirational and i just wanted to say that your content is amazing.
@@CandyMan2001recommend a better course for him if you’re able
Are you up to date on the latest A+? It changes, and is still often a minimum requirement for a lot of jobs and entry level IT person will be looking for. Do it, if only to boost your confidence.
@@CandyMan2001 at my current job... the A+ will give me a $5/hr raise lol and the school they are paying for is paying for the test. NET/SEC + are next along with certs in SQL and LINUX. i'm not going in an expert, but i'm not going to say that the A+ is exactly hard... but it's definitely resume worthy
for such a power hungry device, it would make sense to have a thermal sensor beside the connector. That would make it easy for the card to monitor connector conductor temp and shut down if it got too warm. It would reduce fire risk and returns, and provide extra safeguard for the odd occasion when the connector looks fully on, but isn't.
Appreciate the continuing coverage.
WOW! first time seeing a modern case ad with spaces for optical drives!! I do use those.. sometimes. But mostly need space for card readers.
Thank you, Steve! GN has awesome analysis, and testing. Looks like you identified three issue before Nvidia.
Not sure about that - takes a lot of approvals for a big company to publish!
@gamersnexus glad you’re bringing that up. People really think that the big companies aren’t doing any research and just depending on a single TH-camr to research the problem.
working for the FAA, im pleased to have heard the term FOD outside of the workplace lol, though we use it in the context of ( Foreign Object Damage )
Thank you for all of your diligent work.
So Nvidia confirmed your test conclusions, well done!
If Nvidia indeed also came to the same conclusion they're probably scared of the backlash so they waited for Steve instead lmao
Great work Steve and GN. Thanks for keeping these companies honest.
Thanks to this channel I can finally enjoy my 4090 without worries.
OMG thank god I checked mine after this video, mine came out about 30% and was not seated correctly. Thank you Steve, you saved my 4090 from burning and it's heavily overclocked. I had to shine the torch on the connector to see. Make sure you check yours guys.
Thanks again GN.
what are the odds failure rate goes up once 40XX series are more common in OEM builds. A lot of places already have nominal Q/A, I imagine many of the 16-pin connections might pass 'visual Q/A' but then be loose by the time it gets turned on after shipping & setup.
I think the rate will go up when all of the cards that have already sold actually get sold to people who are using them and installed into systems, they are counting total GPUs sold, and we don't even know if all of those have even been made into cards yet or if it's just the number nVidia has sold itself and to partners. A quick look on ebay and stores will show anyone that they aren't all in use as there is stock for sale still so that .04% failure rate is very generously favoring nVidia at the moment I suspect the real number is going to be much much worse once all 125000 cards actually get into use.
Great research fellas!! I am super interested in a right angled connector so I am not putting those cables under constant strain.
i feel like their response is pretty acceptable. they looked into it, agree with the issue, and are being very open with replacements for folks with the issue. for something that is "user error" this is a big deal, since VERY often that ends up NOT being covered under warranty.
I would say that it's spot on, actually. They acknowledged the failures, investigated properly and are honouring the warranty even if it's partially the user's fault. Aside from faster investigation, what more could you even ask for?
@@jpuroila better connecter design ?
@@roxbantv We live in a fallen world, friend. Though we may strive for perfection, we'll never achieve it.
Igor's suggestion makes alot of sense. It makes it so the sense pins become a physical safety mechanism if they are they are positioned so they only make contact when the connector is fully seated.
The sad part is that this is not some novel idea, even in the computer technology industry it's used widely, just look at the male end of a SATA extension cable for a clear example. Professional equipment that needs to insure that there is no inadvertent removal and also that connections are fully seated often use twist lock connections. CPUs use levers to be both visually obvious and clear to novice users that nothing is going to break. Nvidia just chose to go with the cheapest option that is also most similar to other connectors already in use, which has it's own advantages.
@@msytdc1577 So much wrong there. Those are the signal grounds on a SATA connector. There arent any on a PCI power connector. Also nVidia has f all to do with the cable design as Steve said several times - it is an open cable design form the PCI SIG, the same people the design the PCIe spec
The PCI-SIG suggestion that he reported? Agreed! Makes sense for sure.
@@GamersNexus just add a thin marker at the point on the connector which represents a fully seated position and have that marker align with something. Job done, LOL. Cheap, easy and jank af! haha
@@mycosys The point was that connectors with different length contacts/pins are widely in use and readily known; what each connector is used for is irrelevant. But a common configuration is that the incoming device is grounded first, data lines connected second, and power applied third, in rapid succession as the device is inserted. Also, Nvidia didn't create the cable standard, and no one is suggesting they did,. What they did do was choose to use it instead of a different standard or a proprietary solution if no standards based one already existed. That's on them.
Great journalism like always ! 10 / 10
Thanks to Gamer Nexus for ur time to investigate even show demo how to proper install... i notice my RTX4090 also not properly plugged.. Tips u must check botttom of cable after install and use force to properly plug.. this is definity design flaw.
When you hear "50 cards" it doesn't sound like a lot, until you realise they're about $2,000 each.
And it's even crazier when you realise they've sold 100,000 of them, in a month. That's like $200 million.
Crazy numbers, especially for cards at this price..
Except they don't pocket anywhere near that.
@@Wylie288 Yea I know, I'm just talking about the numbers of cards and the amount of money.
wearing that rib cage shirt today that thing is great. Actually so comfy nice job on that!
Thank you!
Steve and GN must be a huge asset to Nvidia (among other companies). I'm sure GN already saved them several million dollars by now.
If anything, idiots who can't plug in a cable cost nV a lot of money. GN makes money creating content
@@patrickbateman7444 let's call them newcomers to the pc gaming space instead 😉 this can happen to anyone
Say it with me... Not everything is a conspiracy!
@@patrickbateman7444 Somehow people didn't had a problem with standard 8 pin. And considering most 4090 buyers are enthusiasts, they know how to connect a cable
@@wadimek116 99.96% of them did, yes. The other 0.04% were idiots. It's not that hard to understand. And yes, standard 8 pin burned up too on occasion.
Grate job guys, keep up with good work :]
Great info guys! Looking forward to a ‘CableMods’ 90deg 12VHPWR 4090 cable review once released?? 🤞
always cool when the gn team keep communicating with manufacturers. feels like entering a hidden world.
It’s extremely hard to tell whether it’s fully plugged in with such a small connector buried in gigantic heat sink. And there is no clear clicking feedback.
Yeah, I can see that the interface between the 2 connector faces is buried in a shroud. It's got to be very loose before you see that it's not fully seated
Having the connector not click during installation is one thing but having HALF of it not plugged in is pretty easy to spot in my opinion.
Always on point with your videos, thanks for clarifying this issue.
Good work guys all of you involved well done. ❤👍
From my understanding of this whole cable problem it's not an nvidia issue but the cable standard that was voted on that needs to be tweaked in order for it not to become a fire hazard. New technology is never 100% at launch and will need to be fixed overtime. This reminds me of the Wiimote's wrist straps.
Well... this kind of design issue is well known in automotive connector industry.
The connectors have a maximum insertion force specified, minimum audible and tactile feedback, cleanliness and different clip / lock types for how secure the connection has to be.... so about everything that is missing in this standard.
Critical power connectors also have contact features to detect not properly seated connectors - like the proposed change to the 12VHPWR standard
Huff that copium bro @ orlando
Why is this "new technology"? High-amp power connectors are a very old and very well-known technology. This issue has been solved for a hundred years.
Even just using something simple as a positively latching connector (i.e. one where the latch pulls the connector in, like the old Centronics printer plug) would help. It's not magic.
nvidia is a member of PCI-SIG, so they had plenty of input into the design
It's amazing how good these companies actually are at damage control like this. Those clips are usually fine but over the years when these get filled with dust and the clip gets brittle or works loose in a few years due to thermal cycles.
Wow, thanks for this info and video. Very appreciated 👍🏻
Thanks guys. Makes me feel better. I have a brand new system I didn’t turn on with. 4090 because I didn’t want to risk it until known.
125k sold 4090 is really impressive in such a short time. Looks like the huge price hike didn't really bothered their customer base.
Huge price hike? The aib cards msrp is cheaper than the 3090 was at launch
this is the card professionals in the industry will be buying to increase productivity.
17:05 "I promise you no one in this audience has that level of strength"
Audience: 😐
thanks GN for the research! :)
This is good info for an idiot like me who ordered a card with this exact plug. Thanks to all the real ones who had this issues originally. You've earned a new subscriber my man
Wow, GN figured the problem out at basically the same time as Nvidia. Truly impressive!
NV just waited for them otherwise internet would burn if they stated it's user error.
So Jensen is not gonna release a run of T-shirts with "Trust me bro"" written on it?
lol
or "This is still fine" T-shirts.
😂😂😂
The amount of work is insane!
You guys are awesome at exposing these big companies and making it better for us, the end users.
Thank you!
Steve, BIG THANKS to your team and to you personally!
I beleive Ngreedia would've taken even more time to state smth regarding the issue if it wasn't for your great investigation and great work.
Thank you so much, guys!
I hope after this particular guide of yours (13:55) we'll never hear of another "melting" accident.
ps: I'm still wondering why no melting involving the 3090ti cards: the same stupid connector, similar power consumption... not a single "meltig" case.
15:08
connector is fully clicked in, but there is still gap 0.5mm or so. So you clicked it fully and then start managing gabels and gap is still there. Connection is still bad. CONSTRUCTION is still bad and will be melted with time
I get that melting cables issue is a real issue. But the issue I see with this all is the fact I have never heard of any other socketed cables from other GTX or RTX series melting like this. What an odd turn around from the days of lower TDPs.
It's not the TDP itself that's the problem, it's the fact that much power is being pushed through that's small of a connector, so if there is problems it exaggerates them so much more
The older cables were a lot easier to plug in all the way and had ridiculously high safety margins.
I'd also wager that the early concerns about the cable being somewhat delicate led to a positive feedback loop. People would try to be gentle with their cables and fail to exert enough force to plug it in all the way. The cable would then melt, causing the user to warn others to be gentle with their cables. Rinse and repeat.
@@tomhsia4354 arguably correct safety margins
Oh, it did happen. I had a Sapphire Radeon HD 7950 that sufferred a melting connector but on the power supply side (Corsair power supply). It has occurred before but never as frequent as this.
the 3090 also melted however it was not as large of a problem as the 4090
I am amazed the length steve and the team went through to test the actual cause , that cable scanning , xrays was some next level s#it , really impressive
Other way to check connector after install is to take close up photo. You can easily see there if it has space between connector and socket and it's also nice document that how it was installed if there was problem.
Least modifying fix could be a 12vh plug that has yellow plugs/fingers/pins so if it isn’t plugged in all the way the yellow indication is seen. Might create a more expensive cable though as then it would need double molding or some other more complex manufacturing process. Thanks for keeping up with this!
I have an AMD GPU so I'm not going to comment about the main subject of this video. However, the sponsor, Fractal Pop cases. I really like them. I really want one. But you have to pay extra for USB C??? NZXT H510i (which includes RGB and fan controller) is the same price and it DOES include USB C...
Xbox One launched in 2013 with Wi-Fi 5.
PS4 launched in 2013 with Wi-Fi 4.
Companies don't always use logic.
12:00 What about just including a thermister in there somewhere to detect an overheating connector? They do this with AC connectors on some high wattage products. (But most would have never heard of it, as it's molded in.) My car's portable AC charger does this.
Thanks once again. Very informative.
Here's a thought, what about adding thumbscrews instead of a latch to properly torque down the connector into a proper seat instead of that flimsy latch?
Like a VGA-style plug!
Video graphics array plug for a video graphics adapter card.
Or you could just plug it in all the way...
GN deserves a big "thanks Steve" for investigating the problem with the 12VHPWR connectors.
I hope Nvidia will follow up on foreign object debris, and create a better latching mechanism and/or better sense pin safety.
The problem have always been user error. And also, GN isn't just Steve.
Thanks for making great deep info videos.
This " hard shell fully latched" issue is experienced in many industries. Especially in automotive where we are also chasing connector sealing issues from the hostile environments. Also consider , as you have found as most likely user assembly issue, the automobile assembler technicians that assemble thousands of connectors per day with 99.9% success and provide a product lifetime integrity is nothing short of a miracle. Consider the space shuttle electrical connectors. The relationship of hard shell connectors and conductive pins to successfully achieve the primary function of providing voltage and current physical interrupts for user assembly and service, is a complicated science. It really is just good holistic design engineering. All scenarios must be considered. It appears with this issue, someone under valued customer assembly issues. In the big scheme of things, if it is electrical, we all are just chasing Ohm's law. In this case, resistance = heat. Good work GN.