• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

RTX 5090 FE Molten 12VHPWR

Ikasu said:
AC to DC...would have to basically build in a PSU to it. Would add to cost, electrical noise, heat, etc. I'm no electrician or engineer, but just using the noggin for some good ol' common sense. I mean honestly it is a solution...But there are other solutions that are easier...Such as sticking with 8 pin PCI-E -_-...or designing something that isn't as dumb and has barely any safety tolerance at these power level. You'd think they'd at least incorporate some power balancing into the gpu, but nope....Nvidia dropped the ball last gen, and was too bull headed and stubborn to change it. They've single handedly made PSU purchases a fiasco trying to ensure a unit has 12v-2x6, as some manufacturers have not updated their ATX spec listings, and are using the new connector without labeling it. Others label 12v-2x6 on the PSU, but not on documentation, still stating ATX 3.0...Then finally the properly listed ones, which were not many models, but slowly growing.
Click to expand...

well what i'm talking about would be a brick of some sort, not doing ac-dc on the card. i'm sure they could find somewhere in the $2000 budget for it, people are getting ripped off as it is. or just sell it separately and it would be something interchangeable you could use on your next card if you upgrade? and then like someone else was saying they could use 24v-48v if that would be a better solution which isn't available right now from pc power supplies. and it would only be required on these uber high-end cards anyway and then just leave the mid-range cards using standard 8 pin pcie that's worked for decades without issue. i don't know, just throwing out ideas.

i mean amd's still using 8pin pcie on their high end cards and it's not been a problem. but nvidia's trying to use those tiny pcbs on those over-engineered founders editions, and for what? how much money could they cut the price by building a conventional style card? the board partners are getting better thermals with standard coolers anyway. it doesn't really make any sense to me, just gives them a reason to charge two grand for the cards "can't you see how hard it was to put together" yeah we see it, and we don't care! give us a better price and a reason to upgrade.
 
Lakados said:
Nvidia is trying to find a PC equivalent to their SXM modules, and it's not working...

A new power delivery method is needed but I sort of want the AIB's to be the ones to design it, as it's their ass if it goes wrong as the bulk of the machine will be handled by their RMA departments.
They have a strong incentive to get it right out the gate and make it work across everything.
Click to expand...
Pure conjecture.
 
Lakados said:
Nvidia is trying to find a PC equivalent to their SXM modules, and it's not working...

A new power delivery method is needed but I sort of want the AIB's to be the ones to design it, as it's their ass if it goes wrong as the bulk of the machine will be handled by their RMA departments.
They have a strong incentive to get it right out the gate and make it work across everything.
Click to expand...
I forgot to say that as a high as fuck MF I did enjoy this post as I assume you are smoking canuckistan weed while posting,
 
DooKey said:
Pure conjecture.
Click to expand...
Maybe maybe not, but the PCB layouts between their SXM cards and the 5090 are pretty close. Which makes some sense as they are made in the same facility.

We can all agree that this 12v-2x6 connector is still no good even if it is an “improvement” over the 12vhpwr one.

And now we have ASUS, and MSI both working on their own proprietary power delivery designs for their GPU’s that pair with their boards.

I would rather not have my choices be limited to 4 incompatible proprietary specs and 1 garbage one.
 
DooKey said:
I forgot to say that as a high as fuck MF I did enjoy this post as I assume you are smoking canuckistan weed while posting,
Click to expand...
Poutine and a pint while processing paperwork.

I’m eyeballing a 9070 hard though… there is a pretty decent chance that’s happening sooner than not.

I just don’t want to deal with that connector for some upcoming builds even if I’m more than confident in my ability to correctly connect it.
 
mullet said:
In case nobody has seen the fried connector replaced. I love the way this guy talks.


View: https://www.youtube.com/watch?v=rpQnXKMwmpg
Click to expand...


If anyone was wondering how much they charge, their FAQ includes this:

From: https://northridgefix.com/faqs/

Are you accepting GPU repairs?

Yes, we are currently accepting GPU repairs again. Please keep in mind that there is a 50/50 chance of successful repair. Estimated cost to fix these devices ranges from $295-$495.
Click to expand...

I really hope that range is for more speculative troubleshooting and repair where the card just doesn't work, not a simple de/resoldier repair like shown above. Especially since one commenter reported seeing "There were racks on racks of 4090s with melted plugs in his shop." so plug fixes are apparently being done in large enough numbers that work space setup or buying tools shouldn't be a factor.
 
Another issue I just thought about that is glaringly obvious is that when I was a mechanic for years when I was a kid, there is a thing called slide resistance so that any male to female connector has to have a certain tension when sliding the connector together to elevate resistance/heat. Now if that connector is heating up on a regular basis the female connector will loose it's spring tension on the male connector exacerbating the issue making the connection loose it's tight grip on the male connector, making even more heat, until the snowball effect causes total meltdown. Not trying to beat a dead horse but I feel that is 100% what is happening.
 
Looking at this (modified?) Buildzoid slide I saw someone post, I wonder if the design on the Asus Matrix 4090 and Astral 5090 shown here may be some evidence suggesting that Asus may have designed the power regulation/monitoring in a way to justify at least a tiny bit of the excessive price of the Astral? As the Igor's lab link discussed , it seems that the Nvidia demands on the design has made things progressively worse from the well made 3090 layout and some AIBs doing even better, down to the 4090 and 5090. As he discussed with MSI, perhaps there is evidence that AIBs from Asus and perhaps others could at least mitigate some of the risk of Nvidia's designs with their solutions. If this is accurate, I'd be interested to compare the Astral's power delivery system to MSI's Vanguard (I think that's the parallel high end card?) ,a s well as other AIBs higher end card variants first, and then move down the lines (ie Astral vs TUF) to see if some manufacturers and card models have worthwhile mitigations as opposed to the FE.
 

Attachments

  • 1739451675567754.png
    1739451675567754.png
    1.1 MB · Views: 0
RanceJustice said:
Looking at this (modified?) Buildzoid slide I saw someone post, I wonder if the design on the Asus Matrix 4090 and Astral 5090 shown here may be some evidence suggesting that Asus may have designed the power regulation/monitoring in a way to justify at least a tiny bit of the excessive price of the Astral? As the Igor's lab link discussed , it seems that the Nvidia demands on the design has made things progressively worse from the well made 3090 layout and some AIBs doing even better, down to the 4090 and 5090. As he discussed with MSI, perhaps there is evidence that AIBs from Asus and perhaps others could at least mitigate some of the risk of Nvidia's designs with their solutions. If this is accurate, I'd be interested to compare the Astral's power delivery system to MSI's Vanguard (I think that's the parallel high end card?) ,a s well as other AIBs higher end card variants first, and then move down the lines (ie Astral vs TUF) to see if some manufacturers and card models have worthwhile mitigations as opposed to the FE.
Click to expand...

It at least can notify you of amperage per pin via the GPU Tweak tool. So if you are having uneven power draw, you would at least know of it and be aware which is helpful. Still doesn't excuse the travesty that is the 4090/5090 power/connector issue. But it's at least something.
 
mullet said:
I watch these repair guys doing repairs all the time, big factor on Nvidia is torn pins behind the gpu chip.


View: https://youtu.be/SdymB1vLdF8?t=197
Click to expand...


I think he does impressive work, but I quite dislike his titles. Why is that going to happen to 50% of them? He always finds something to say is going to happen to a "lot of GPUs" (like blowing canned air into the fans, the cracked retention clip which apparently does have traces, etc), but not really why. Is it just the heavy heatsinks? I guess if so, I'm lucky I decided to get the Suprim X Liquid, because the heatsink on the actual core (and the PCB in general) is going to be incredibly light compared to most of the monstrosities on air cooled 4090s. But it sucks that this is even an issue, and it's really stupid.


One other thing that I don't think I have seen mentioned in this topic is 12VHPWR cable bending. According to most manufacturer specifications I have seen, you're supposed to give the 4090 12VHPWR connector a good bit of cable length before you even BEGIN to bend it. But how the fuck do you do that? I don't think I've seen very many people that actually listened to that instruction, but it's not due to lack of trying it's because it's damn near impossible in a vast majority of cases due to how wide the damn card is. Part of the reason I got this CTE C750 case is because I noticed the extra width would barely allow me to keep the card in spec, and then on top of that the orientation would allow me to tie the cable to the AIO tubing for extra support.

The 5090 will be better in this regard because its connector is slanted, but only "better".

This whole thing is just a shitshow....
 
StoleMyOwnCar said:
I think he does impressive work, but I quite dislike his titles. Why is that going to happen to 50% of them? He always finds something to say is going to happen to a "lot of GPUs" (like blowing canned air into the fans, the cracked retention clip which apparently does have traces, etc), but not really why. Is it just the heavy heatsinks? I guess if so, I'm lucky I decided to get the Suprim X Liquid, because the heatsink on the actual core (and the PCB in general) is going to be incredibly light compared to most of the monstrosities on air cooled 4090s. But it sucks that this is even an issue, and it's really stupid.


One other thing that I don't think I have seen mentioned in this topic is 12VHPWR cable bending. According to most manufacturer specifications I have seen, you're supposed to give the 4090 12VHPWR connector a good bit of cable length before you even BEGIN to bend it. But how the fuck do you do that? I don't think I've seen very many people that actually listened to that instruction, but it's not due to lack of trying it's because it's damn near impossible in a vast majority of cases due to how wide the damn card is. Part of the reason I got this CTE C750 case is because I noticed the extra width would barely allow me to keep the card in spec, and then on top of that the orientation would allow me to tie the cable to the AIO tubing for extra support.

The 5090 will be better in this regard because its connector is slanted, but only "better".

This whole thing is just a shitshow....
Click to expand...


I think these 2 repairmen of those 2 youtube channels have forgotten more about electronics and how they work then JayzTwoCents and Gamers Nexus combined understand.

I think youz guyz should watch more of the repair videos, you can learn a ton.
 
mullet said:
I think these 2 repairmen of those 2 youtube channels have forgotten more about electronics and how they work then JayzTwoCents and Gamers Nexus combined understand.

I think youz guyz should watch more of the repair videos, you can learn a ton.
Click to expand...

My issue is the hyperbole. I'm sitting back and just watching and waiting because right now there's a ton of "the sky is falling" stuff going on. People making claims that "half the cards will fail" is just being overly dramatic to get people to watch videos. But, people walk away and tell others that "half the cards are failing!!!!". Is there a problem? If I look at the raw facts I think there is a deficiency in the design. Maybe the deficiency causes problems in 1 in 1000 cards, or 1 in 1,000,000 cards. We don't have data yet to make that call. To be frank, people who run out and buy the best of the best on the day of release are taking a risk. They know they're taking a risk. Most of us sit back and wait until the dust settles because it is so much less stressful.
 
Ididar said:
My issue is the hyperbole. I'm sitting back and just watching and waiting because right now there's a ton of "the sky is falling" stuff going on. People making claims that "half the cards will fail" is just being overly dramatic to get people to watch videos. But, people walk away and tell others that "half the cards are failing!!!!". Is there a problem? If I look at the raw facts I think there is a deficiency in the design. Maybe the deficiency causes problems in 1 in 1000 cards, or 1 in 1,000,000 cards. We don't have data yet to make that call. To be frank, people who run out and buy the best of the best on the day of release are taking a risk. They know they're taking a risk. Most of us sit back and wait until the dust settles because it is so much less stressful.
Click to expand...


If you watch these videos from both of the repair guys, you will say to yourself, I had no idea so many connectors are melting, I had no idea so many of the pcie connectors are cracking, I had no idea the bloke pins under the GPU processor which there is NO fix.
Why do I say this is because only the repair guys are going to know this, it's not like there is a website that everyone knows about to log their GPU failure and what is was, if there was then we would all know about it.

Do you think Nvidia is going to post all of the failure rates and what was the cause? NO

Is there come click bait on the titles of youtube videos? YES
 
Ididar said:
To be frank, people who run out and buy the best of the best on the day of release are taking a risk. They know they're taking a risk. Most of us sit back and wait until the dust settles because it is so much less stressful.
Click to expand...
I learned my lesson from the 4000 series cards. Sit back and wait this time, or just maybe not buy anything at all.
 
If nothing was done to correct the load balancing issue on the 4090, we can assume there will be a higher rate of incident on a card that consumes far more wattage. It's so close to spec, you need that much more of a "perfect" contact. That much less resistance differential to create heat and possible failure. I want a 5090, but some kind of active load balancing is needed. Looks like we'll only get it PSU side, or with a recall of the product.
 
2 12v-2x6 limited to 450 watts will offer a lot more margin and be physically smaller than 4 8 pins. I feel like that would be a better option, and allows up to 900 watts on a single card. Kinda ridiculous if/when we ever get to 900 watts...
 
What's happening to peoples' electrical engineering skills? Maybe new engineers are passing college using AI 😂 ! I used to play with electronics/electrical toys when I was a kid, and the way I used to connect wire leads to batteries was duct tape (Didn't have access to soldering iron back then). I was really pissed that not all battery power will go through my circuits because I was under the impression that if it touches, then it should be okay. Applying little force over the duct tape in both battery terminals will cause light bulbs to glow brighter. That's when I learned bad contact is the worst thing that can happen to electrical circuits.
 
Those power ratings are based on splitting load *equally* across all pins which clearly is not happening. For those of you old enough to remember when households had Edison fuses, the trick of putting a penny under a fuse that kept blowing comes to mind. This is a good way to burn down your house!

RE: bricks...

This comes to mind. RV power converters. Years ago these were linear devices (used step down iron core transformer, full wave bridge rectifier and some filter caps). Now they are switchers just like a PC power supply:

6CsBCOt5UOrBvKK4EJ4MHr5PjbGz4ZtOt6p92Fp09wnJ7Eko3A.jpg
 
sram said:
What's happening to peoples' electrical engineering skills? Maybe new engineers are passing college using AI 😂 ! I used to play with electronics/electrical toys when I was a kid, and the way I used to connect wire leads to batteries was duct tape (Didn't have access to soldering iron back then). I was really pissed that not all battery power will go through my circuits because I was under the impression that if it touches, then it should be okay. Applying little force over the duct tape in both battery terminals will cause light bulbs to glow brighter. That's when I learned bad contact is the worst thing that can happen to electrical circuits.
Click to expand...

People love to over complicate simple stuff. If you are pushing the boundaries of wire and connector and increasing the failure at the same time. It's pretty simple to fix, increase wire size and connector size. Solved.
 
mullet said:
Another issue I just thought about that is glaringly obvious is that when I was a mechanic for years when I was a kid, there is a thing called slide resistance so that any male to female connector has to have a certain tension when sliding the connector together to elevate resistance/heat. Now if that connector is heating up on a regular basis the female connector will loose it's spring tension on the male connector exacerbating the issue making the connection loose it's tight grip on the male connector, making even more heat, until the snowball effect causes total meltdown. Not trying to beat a dead horse but I feel that is 100% what is happening.
Click to expand...
Molex connectors were never made for this kind of current. And yes, the alleged covered area can "pass" higher current levels than the cross sectional area of the pins. A bad connection creating more resistance certainly will exacerbate this problem tenfold. Simply overloading a single conductor, OTOH, will cause severe overheating off the bat. By the time one smells burning plastic, things are already messed up. But people play on. I could not tell you how many times people continue to bump their sub boxes in their cars long after they smell voice coil varnish cooking off. They'll give it a break for a while then proceed to redline the crap out of the amp.
 
mullet said:
People love to over complicate simple stuff. If you are pushing the boundaries of wire and connector and increasing the failure at the same time. It's pretty simple to fix, increase wire size and connector size. Solved.
Click to expand...

NightReaver said:
"Noooooooo but muh aesthetics!" People would rather risk fires/ electrical issues instead lol.
Click to expand...
The connectors are the sizes they are, but they are the same connectors that are in the 8-pin connectors we currently use, those connectors are as big as they are getting and any changes there will change the size and shape of the 12V connector itself, so then we are looking at a whole new connector which is needed.
I am going to call the 12VHPWR / 12V-2x6 connector dead, consumer faith in it is gone, and in its incredibly short lifespan, it has just been the center of one controversy after the next, ultimately it will end up being a short-lived fad.

I'm sure that PSU manufacturers will figure out how to build a modular connector to maintain the PSU's compatibility with whatever replaces it, in the (hopefully near) future.

OEMs and SIs have a big problem here, this is too easy to reproduce, what if they wiggle loose during shipping? Will they require the users to inspect the cables upon delivery? Would that mean the user having inspected the cable assumes responsibility for them should the worst occur and is no longer covered by warranty?

These connectors need to be retired, and something more robust and user-friendly needs to take it's place.
 
Tsumi said:
2 12v-2x6 limited to 450 watts will offer a lot more margin and be physically smaller than 4 8 pins. I feel like that would be a better option, and allows up to 900 watts on a single card. Kinda ridiculous if/when we ever get to 900 watts...
Click to expand...
975 watts. Can't forget the PCIe based power. Also crazy to think of Asus's BtF based power delivery that would push all that wattage into the motherboard itself for use by a video card.

I do agree that if a higher power rated cable over 8 pin was warranted, having a ~50% overhead safety margin is warranted.

cpufrost said:
Molex connectors were never made for this kind of current. And yes, the alleged covered area can "pass" higher current levels than the cross sectional area of the pins. A bad connection creating more resistance certainly will exacerbate this problem tenfold. Simply overloading a single conductor, OTOH, will cause severe overheating off the bat. By the time one smells burning plastic, things are already messed up. But people play on. I could not tell you how many times people continue to bump their sub boxes in their cars long after they smell voice coil varnish cooking off. They'll give it a break for a while then proceed to redline the crap out of the amp.
Click to expand...

Worse would be if Nvidia's 12V-2x6 pins were to become as loose a Molex's pins tend to be after merely half-a-dozen connections.
 
Ididar said:
My issue is the hyperbole. I'm sitting back and just watching and waiting because right now there's a ton of "the sky is falling" stuff going on. People making claims that "half the cards will fail" is just being overly dramatic to get people to watch videos. But, people walk away and tell others that "half the cards are failing!!!!". Is there a problem? If I look at the raw facts I think there is a deficiency in the design. Maybe the deficiency causes problems in 1 in 1000 cards, or 1 in 1,000,000 cards. We don't have data yet to make that call. To be frank, people who run out and buy the best of the best on the day of release are taking a risk. They know they're taking a risk. Most of us sit back and wait until the dust settles because it is so much less stressful.
Click to expand...

I don't think anyone is saying the sky is falling, but it's a definite issue. There are very few 5000 series cards out there and we have had quite a few reports of issues. But I myself do not like the design, reminds me of the weak wiring they tried to use on blower motors on vehicles and they had the same problems, melted wiring and connecters. Push a 5090 card and I feel most will have some type of issue. 5080 and down are less likely unless they overclock the crap out of them. Just not wise to push a design right up to the max.
 
Ididar said:
To be frank, people who run out and buy the best of the best on the day of release are taking a risk. They know they're taking a risk. Most of us sit back and wait until the dust settles because it is so much less stressful.
Click to expand...

sure early adopters are taking a risk but not the kind of risk where the product will catch on fire...early adopters pay more, take the risk of bugs and performance issues but not that their product is going to blow up
 
polonyc2 said:
sure early adopters are taking a risk but not the kind of risk where the product will catch on fire...early adopters pay more, take the risk of bugs and performance issues but not that their product is going to blow up
Click to expand...
This^. Nobody should have to worry about a hazardous risk to themselves or to their home.
 
polonyc2 said:
sure early adopters are taking a risk but not the kind of risk where the product will catch on fire...early adopters pay more, take the risk of bugs and performance issues but not that their product is going to blow up
Click to expand...
it's no longer the bleeding edge of technology but the burning edge...
 
DPI said:
Nobody is actually disputing this. But concern trolling is a thing, especially when there's a financial incentive (youtubers, tech sites).
Click to expand...
when you got guys making a living fixing melted graphics cards i don't think it's necessarily trolling, financial incentive well yeah, and think about it, posting youtube vids is advertising that pays YOU these days instead of the other way around. like they say don't hate the player hate the game or, if you're an nvidia customer, hate the flame
 
I think the repair dudes are great, you can find out about the brand issues, learn something, they even sell the tools they use, heck maybe some on this forum will end up fixing cards. I like the idea of [H] member repairing cards for us with senior discounts of course.
 
You must log in or register to reply here.
Back
Top