Or even the cost of a repair, which is usually around $400, not to mention being without your GPU for several weeks.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
RTX 5090 FE Molten 12VHPWR
- Thread starter erek
- Start date
Put some Frank's Hot Sauce on it.
jspinks231
Weaksauce
- Joined
- Aug 31, 2026
- Messages
- 86
It’s hard to know what to do or say with a company that can’t make a 5-10 dollar plug so no me with nvidia..I don’t trust and my already shill non deer non Andyson Apevia gold galaxy 650w more “in house” unit..I’m not trying to burn mine with in 5 years..I’d like my units to run 7-8years..
But yeah good stuff been countless reports of them burning..I’ll be real conservative with you and just have my 1080. Vega56 hbm2 model..I don’t trust that plug on a 200w one if it burn the whole corner out…you are right having way wrong 400-500w transients that high…but really a lot people are putting some junk 80bronze Andyson units on them…or something else just as bad as deer and will Tripp in 3-5 years.
But yeah good stuff been countless reports of them burning..I’ll be real conservative with you and just have my 1080. Vega56 hbm2 model..I don’t trust that plug on a 200w one if it burn the whole corner out…you are right having way wrong 400-500w transients that high…but really a lot people are putting some junk 80bronze Andyson units on them…or something else just as bad as deer and will Tripp in 3-5 years.
Yep the good thing about the wp2 after having one and it finding my cable was bad I feel like it will definitely let me know again if it goes bad before something burns. But more importantly i feel like the wp2 to GPU side should stay "healthy" longer and hopefully the cable end is what keep falling
- Joined
- Dec 19, 2005
- Messages
- 19,153
- Joined
- Aug 27, 2025
- Messages
- 644
What is you tally so far?
33 pages and how many cards do you have found?
5090's: ?
5080's: ?
5070's: ?
4090's: ?
4080's: ?
Others.
Because you do keep track, not just post buthurt right?
(My 4090 from 2022 is doing fine btw)
- Joined
- Dec 19, 2005
- Messages
- 19,153
33 pages and you are still eaither retarded or trolling as I had a problem with a 5090 cable and so did more then a few here. Doesn't it get old be a troll? Do you really get that big of a dopamine hit form being a dick?
Tell you what ues an AI and make a list of this thread of everyone that had problems and get back to us with the numbers it would be interesting wouldn't?
people feeding a troll are just as annoying as the troll itself.
Block and move on with your life.
I am not sure, according to AI:
HWinfo forum.
***
I updated HWinfo to the latest version, and that value is no longer there but there is 0.870v-12.228v:
Last edited:
applegrcoug
Gawd
- Joined
- Aug 28, 2021
- Messages
- 946
Well, not that I have a 5090, because I don't....
But I do have a 4090 and a 5070ti and the way these things are going up in value it is making me nervous. I've been thinking about getting a wireview pro II for at least the 4090 considering that is where the molten cables originated and I was messing around with comfyui and now with the gpu unlock, that thing will just suck electricity.
However, is it even warranted for a 5070ti and the reduced power draw?
But I do have a 4090 and a 5070ti and the way these things are going up in value it is making me nervous. I've been thinking about getting a wireview pro II for at least the 4090 considering that is where the molten cables originated and I was messing around with comfyui and now with the gpu unlock, that thing will just suck electricity.
However, is it even warranted for a 5070ti and the reduced power draw?
- Joined
- Mar 18, 2010
- Messages
- 14,501
Given that it has melted at loads of 300 watts, I would say yes.
I have no idea what that is, but if it is operating normally it is more likely an error in the software. I've noticed HWINFO does not get all my X570 MSI motherboard sensors correct.
Just taking a WAG, it could be voltage that represents something else like current and perhaps has a range from 0V to 12V corresponding to minimum and maximum current. Sounds like the HWINFO developer found it, but doesn't have docs on what it represents.
Looking at the way these are climbing, I think I would.
Another reason to consider it is that one may be more tempted to push a 5070TI because of its lower power rating with the desire to extract 5080 like performance out of it. I know I always firewalled my sub 80 series cards when I had them.
Is $200 worth not having to worry about it to you? It was to me and I slapped one on the 5090 and it instantly identified I had a bad cable. For the first time sense I had the thing I feel I can enjoy gaming on it without worrying in the back of my mind if it's okay or not. Worth every dam penny x10 in my book.
I finally tossed my WP2 on my 5090, had to do some HS swaps with my other box to get the offset D15 to make room for it, which i conveniently didn't think about when ordering. Fun little thing, definitely like it. I still run the 5090 at 80% power limit, but that's more because it makes my room too hot under full load (Generally owould only hit over 450-500 watts when I was on full PT/RT stuff anyway, so normal gaming at 4K maxed is still fine at that PL for most games with just RT and no PT). Probably spend way too much time thinking about things to tie into the sensors lol.
One saved my 5090 the same way. Got another for my 4090 but it is fine (for now).
I ran 80% before getting the WVP2 (not that it would have stopped it from eventually melting), but now that I've got one (and replaced the bad cable) I run it at 100% knowing that it'll alarm on me if there's a problem well before it melts.
I still run it at 500w as I'm not sure they will be a 6090 at the rate things are going and I'll rather baby it as you really don't give up much fps at 450-500w and if I need more power I can crank it up. But I don't forsee to many developers pushing the envelope anytime soon.
The price and availability on computer parts has basically made me detune everything in the hopes everything lasts till this dry spell ends or I die. We are definitely living in some interesting times or maybe I'm just getting to old man age and the excitement of computers finally wore off I don't know
mullet
2[H]4U
- Joined
- Aug 19, 2004
- Messages
- 3,512
The 6090 should have it's own power supply.
- Joined
- Dec 19, 2005
- Messages
- 19,153
- Joined
- Dec 19, 2005
- Messages
- 19,153
applegrcoug
Gawd
- Joined
- Aug 28, 2021
- Messages
- 946
for the 4090, yes. That's what i said my plan was provided they ain't a hunk o' junk, which it seems they aren't.
Just unsure if it is warranted on a 5070ti. That is my conundrum.
- Joined
- Dec 19, 2005
- Messages
- 19,153
as someone mentioned above the 4090, 5090 (& maybe 5080 too) seem to have too much juice flowing thru them
maybe nvidia needs to focus more on the power efficiency curve, for next gen
Nvidia GPUs as of late are general more power efficient that equivalent Radeon GPUs. The problem is they are not implementing load balancing on the card so that on or two pins don't draw more amps than spec. 8-pin PCI-E are less likely to melt because they have a higher safety factor (1.8) before they hit hardware limits that lead to melting. The safety factor for 12v-2x6 and the 12HPWR connector is about 1.2, so much lower than the venerable 8-pin PCI-E connector. I would rather have a larger cable (less resistance than multiple tiny cables) such as a pair of fine stranded 12 AWG wires than six 16 AWG wire. 12 AWG fine stranded wire isn't hard to bend if the wire jacket is modest.
- Joined
- Apr 25, 2001
- Messages
- 16,289
According to my WVP2 all but one wire are pulling the same amps and the other is .1amp higher. This is across all load ranges. Looking good so far.
Last edited:
- Joined
- Dec 19, 2005
- Messages
- 19,153
- Joined
- Aug 27, 2025
- Messages
- 644
That was a zero answer, figured you are not interested in the data.
- Joined
- Aug 27, 2025
- Messages
- 644
33 pages mostly of posts not containd dead card, but only fluff, just like your post
Curious that no one is keep a tally, it is almost like the numbers do not matter to a certain segment.
hititnquitit
2[H]4U
- Joined
- Apr 16, 2018
- Messages
- 3,146
So you come here to whine about other people not doing what you're to lazy to do yourself.
That'll show em.
- Joined
- Dec 24, 2008
- Messages
- 2,318
What was the default behavior on previous cards that had individual or dual 12V pins connecting to a single VRM? Let's say the wire(s) had abnormally high resistance, so then Im assuming the VRM would shutdown from undervoltage. Was there some message to the user on the next boot that the +12V input system failed, or was it up to the user to guess what the issue might be? Just curious.
I know on my Titan XP that not plugging in the PCIe 6 pin connector in addition to the PCIe 8 pin resulted in the driver not loading with some error in device manager. It did not outright say it was a power issue. I found it when I went over the hardware and found the connector not seated. Plugged it in fully, powered up, and the driver loaded.
I believe at some time I failed to plug in the pairs of PCIe 8 pin connectors on a pair of 1080TI FTW3 cards and the result was no video and I don't think the system even booted. No messages other than the no workie.
I don't remember which card it was, as it was a long time ago, you would get a message on boot to plug in the power and the system would just hang there, you couldn't even get into BIOS. I was running a CRT monitor at the time.
The voltage regulator has the ability to monitor the 12V it receives as well as current if it has shunts. As the 12V falls, more current is required to meet the wattage requirements of the card. It seems logical to me that you would set the voltage cut off at the point at which the current would overload the connectors or even the MOSFETs in the regulator itself. This is pretty standard stuff and the feature is usually referred to as an undervoltage lockout.
Well if I remember right I ran a few mid range amd cards with just one cable plugged in mining back in the day and they worked for over a year that way. Pcie was next to impossible to kill. You could split one cable to two plugs for years and years and nothing happened
- Joined
- Dec 24, 2008
- Messages
- 2,318
On a separate note: I'm still not sure how multiphase VRMs are typically setup for load balancing. Balance the
1) input current (high side mosfet)?
2) the output current ( inductor)? (what i've been caling vrm power balancing, probably confusingly so)
I was browsing these 2 Ti documents. I dont know how representative they are of multiphase VRM designs in GPUs, but its implying that the inductor current (low side) is the more important one to balance. So I wonder if that's how motherboard/GPUs are typically setup to load balance.
https://www.ti.com/lit/ta/ssztc61/ssztc61.pdf?ts=1794454084556&ref_url=https%3A%2F%2Fcn.bing.com%2F
https://www.ti.com/lit/an/slva882b/slva882b.pdf?ts=1659274643749
edit: fixed mistake where I made assumption that lowside mosfet and inductor current are the same. incorrect assumption.
1) input current (high side mosfet)?
2) the output current ( inductor)? (what i've been caling vrm power balancing, probably confusingly so)
I was browsing these 2 Ti documents. I dont know how representative they are of multiphase VRM designs in GPUs, but its implying that the inductor current (low side) is the more important one to balance. So I wonder if that's how motherboard/GPUs are typically setup to load balance.
https://www.ti.com/lit/ta/ssztc61/ssztc61.pdf?ts=1794454084556&ref_url=https%3A%2F%2Fcn.bing.com%2F
https://www.ti.com/lit/an/slva882b/slva882b.pdf?ts=1659274643749
See figure 4-1, where DCR method measures current from the output current across the inductor
edit: fixed mistake where I made assumption that lowside mosfet and inductor current are the same. incorrect assumption.
Last edited:
- Joined
- Dec 24, 2008
- Messages
- 2,318
Some more:
https://www.infineon.com/assets/row/public/documents/24/49/infineon-xdpe132g5h-g000-datasheet-en.pdf
infineon vrm controller that uses TDA21470 drivers for the phases. Note the current sense connection IOUT on each.
https://www.infineon.com/assets/row/public/documents/24/49/infineon-tda21472-datasheet-en.pdf
https://www.infineon.com/assets/row/public/documents/24/49/infineon-xdpe132g5h-g000-datasheet-en.pdf
infineon vrm controller that uses TDA21470 drivers for the phases. Note the current sense connection IOUT on each.
https://www.infineon.com/assets/row/public/documents/24/49/infineon-tda21472-datasheet-en.pdf
This represents inductor current (low side output)
It's possible they could do both. Having both the input and output currents allows the regulator to also measure/account for it's efficiency. If you really want to get into the nitty-gritty of switching regulator design you are looking into a deep crevasse, albeit an interesting one*.
Power out of the regulator is power in to regulator minus losses. Efficiencies above 85% are generally considered good and getting in the high 90s is excellent. With poor design or application, it is very easy to wind up with efficiencies in the 60s. Efficiency matters because the more efficient, the less self heating of the regulator, and the more power it can deliver.
*Off topic, but about switching power supplies, the LT8203 is an interesting device. It is a flyback switching regulator that you can build isolated power supplies from. The isolation is provided via a transformer and here is its trick - there is no direct feedback from the regulated output. The moment the MOSFET turns off, the output voltage can be sensed on the primary side of the transformer. This eliminates the need for an optocoupler or additional transformer windings to sense the isolated output voltage. "Tricks" like this are all over the place in switching power supply design. The output voltage and its ground are electrically isolated from the power supply input.
- Joined
- Dec 24, 2008
- Messages
- 2,318
From the rest of my posts it's leaning to 2) output current (inductor current) is what is balanced. Not input. Unless you have seen examples otherwise?
There is balancing going on all over the place in switching power supply design.
Load balancing at the input is used to protect the power feed wiring and connectors. That's all it is for.
I'm sure another concern of a multi-phase power supply is balancing the output current across the phases. Most of the power supply output is destined to supply the common core voltage. At some point all the power feeds and core power regulator output converge on a single source that powers the GPU core voltage. Other, lesser supplies, power the DRAM and other stuff on the card.