12VHPWR has a higher number of reported failures than 6/8 pin does despite being much younger

[dirtyferret]

I see you are referring to the exhaustive study and 1 hr 12 minute video from Gamer’s Nexus?

The majority of problems stemmed from angled connectors, and of those the highest number were from 2 brands. I think also the problems were nearly all from aftermarket connectors, not ones that actually came with ATX-compliant PSU’s.

It’s not so much a big deal.

agreed, anything that guy posts I take it with a large heaping of salt. He's gone from publishing excellent reviews on his web site to becoming his own brand and constantly pushing the envelope on narratives with click bait headlines just so he can have content for profit. I get the business end of it and which him well but why people fall for a lot of his stuff is mind numbing.

I don't know.. maybe I am just good with my hands? Even my first try plugging it in was not a problem at all, heard the click and I was set. Once its in, you can pull on it, do whatever.. within reason.

Let's not bring reason and common sense into this discussion. It's pitch fork and torch time baby!/s

 

[dont whant to set it"']

@nguyen the Fury Nano is one that could stretch it's legs to 300W.

 

[nguyen]

@nguyen the Fury Nano is one that could stretch it's legs to 300W.

There are XOC bioses on Nvidia highest end GPU (1080Ti - 2080Ti - 3090) that could push over 300W per 8pin PCIE power, but they are not the standard PCIe specs of 150W per 8pin so it doesn't matter.

The Fury Nano is a sub 225W GPU, so 150W over 8pin + 75W over the PCIe slot.

That remind me of the time when Radeon draw too much power on the PCIe slot and burning the slot LOL.

AMD Radeon RX 480 Reportedly Frying PCIe Slots

Following reports that AMD’s new Radeon RX 480 was exceeding the 75W power draw limit through PCIe, users have posted on the AMD forums to complain that the graphics card has fried their PCIe slots.

www.eteknix.com

 

[evernessince]

Nothing of the sort, huh? The primary focus of rhe video was with angled connectors that aren’t OEM.

The primary focus is 12VHPWR. I don't see your insistence on otherwise, it's in the title. Right angled connectors were talked about for a fraction of a 22 minute section of an hour+ long video.

agreed, anything that guy posts I take it with a large heaping of salt. He's gone from publishing excellent reviews on his web site to becoming his own brand and constantly pushing the envelope on narratives with click bait headlines just so he can have content for profit. I get the business end of it and which him well but why people fall for a lot of his stuff is mind numbing.


Let's not bring reason and common sense into this discussion. It's pitch fork and torch time baby!/s

Reason would be debating the points presented and not making character assassinations that don't counter any points make.

 

[Vincero]

The gold nugget of this video though is this chart illustrating the total number of people who experiences issues with 6/8 pin vs 12VHPWR:

View attachment 366533

Your post title is misleading...

You are either mis-understanding the scope of the graphs shown or deliberately trying to draw an impossible conclusion based on the simple premise of manufacturing over time.

This is based as a percentage. 3.3% of several million of 6-pin/8-pin graphics cards vs 4% of thousands or hundreds of thousands or maybe even a million or so 12VHPWR graphics cards DOES NOT mean that there has been more failures of those within a couple of years vs 15+ years of the existing 6+2pin PCIe power connection existing. I'd hazard a guess that more graphics cards that were manufactured with 6+2pin connectors have been scrapped/recycled than 12VHPWR have been manufactured within the last 2 years.
It's also based on survey data, not actual recording of hardware failures/returns or the result of some intermediate party that has some reason to keep statistics on such a thing (e.g. an insurance company, etc.), so it's not exactly an infallible data set.

Using that same percentage against a sample of 100 of each type would mean that maybe 3 of one type and 4 of the other type would have issues. In any case they would need another 20+ years of both types being manufactured equally before they would even be close of approaching a similar number of overall failures.

I'm not defending the 12VHPWR standard - as soon as I saw it and the power limits designed for it I was like "ok... we'll see how that goes" having already seen failures on the existing 6+2pin system which (as GN itself touched upon) has quite an amount of headroom in its design. To be honest, I was surprised to see so many issues straight out of the gate - I expected it to at least be more of a used / second hand market type of issue.
Also, the simple premise of 'make something better than existed before' has certainly not been the case - if anything 12VHPWR has made the argument for the HPCE + PCIe connector (see link) a much stronger one even if that means backwards compatibility may go out the window or a load of cases with seperate GPU mounts and extension cables need to replacement parts (unless some cards have an auxiliary port).

 

[chrcoluk]

View attachment 366604

This is a photo taken this morning of the CableMod C-Series Pro ModMesh I purchased about a year and a half ago. Only maintenance done is a small drop of DeOxIt D100L on the 12V and ground pins every few months or so.

View attachment 366605

This RTX 4090 is approaching two years of use with zero issues related to power delivery. The card is stock with no BIOS or voltage modifications made.

I don't know what the root cause of the melting connectors is, nor am I going to speculate on it or argue with anyone. I'm just responding to the categoric insistence that it's the fault of CableMod or the fault of the connector, because I'm providing a real-world use case of both of these operating within design specifications.

I did have a melting power connector problem in the past -- it was an ATI All-In-Wonder 9800 Pro where it had one typically used on floppy drives, and it clearly wasn't up to the job.

Poor Steve must be running out of things to create his usual whiny negative content about. Maybe he can find another overheating no-name pre-built at Wal-Mart with poor air flow or go on yet another unhinged rant about Alienware while he's at it. Got tired of his "I'm the smartest guy in the room" schtick years ago.

Yeah as I replied above, the issue cablemod had is with their adaptors, not the cables, they still sell the cables and have done no recall on them. Thanks for posting about the state of your own cable.

Your post title is misleading...

You are either mis-understanding the scope of the graphs shown or deliberately trying to draw an impossible conclusion based on the simple premise of manufacturing over time.

This is based as a percentage. 3.3% of several million of 6-pin/8-pin graphics cards vs 4% of thousands or hundreds of thousands or maybe even a million or so 12VHPWR graphics cards DOES NOT mean that there has been more failures of those within a couple of years vs 15+ years of the existing 6+2pin PCIe power connection existing. I'd hazard a guess that more graphics cards that were manufactured with 6+2pin connectors have been scrapped/recycled than 12VHPWR have been manufactured within the last 2 years.
It's also based on survey data, not actual recording of hardware failures/returns or the result of some intermediate party that has some reason to keep statistics on such a thing (e.g. an insurance company, etc.), so it's not exactly an infallible data set.

Using that same percentage against a sample of 100 of each type would mean that maybe 3 of one type and 4 of the other type would have issues. In any case they would need another 20+ years of both types being manufactured equally before they would even be close of approaching a similar number of overall failures.

I'm not defending the 12VHPWR standard - as soon as I saw it and the power limits designed for it I was like "ok... we'll see how that goes" having already seen failures on the existing 6+2pin system which (as GN itself touched upon) has quite an amount of headroom in its design. To be honest, I was surprised to see so many issues straight out of the gate - I expected it to at least be more of a used / second hand market type of issue.
Also, the simple premise of 'make something better than existed before' has certainly not been the case - if anything 12VHPWR has made the argument for the HPCE + PCIe connector (see link) a much stronger one even if that means backwards compatibility may go out the window or a load of cases with seperate GPU mounts and extension cables need to replacement parts (unless some cards have an auxiliary port).

Not a fan of the HPCE idea, we then become reliant on motherboard quality to provide quality power delivery, and of course boards will become even more expensive, there was really nothing to fix on the standard 8 pin cables, the issue to fix is to stop making 600w GPU's and stop trying to reinvent the wheel on the connectors. Imagine needing a new motherboard instead of a new PSU or new cable to meet a new power delivery standard. Then the slot burns on defective and corner cut boards.

 

[Vya Domus]

This never ending shitshow could have been avoided if the industry simply moved to a redesigned 1x8pin type connector with heavier gauge wiring, etc, and a more modest 350W rating or something like that and if your GPU is extra power hungry you just use 2 of them. But what I bet happened is that Nvidia was dead set on using a single connector on everything, so they can do their idiotic Apple like "clean and simple" design aesthetic that they're clearly trying to copy for a while.

I heard it's not their invention but I got no proof on hand

Yeah sure, it's not their invention, that's why they're putting it in every single consumer product of theirs.

It was commissioned by Nvidia and Dell, later adopted by Intel in the ATX spec, of course it's their invention.

 

[Vincero]

Not a fan of the HPCE idea, we then become reliant on motherboard quality to provide quality power delivery, and of course boards will become even more expensive, there was really nothing to fix on the standard 8 pin cables, the issue to fix is to stop making 600w GPU's and stop trying to reinvent the wheel on the connectors. Imagine needing a new motherboard instead of a new PSU or new cable to meet a new power delivery standard. Then the slot burns on defective and corner cut boards.

I can't say I'm a fan either, for many of those same reasons but also because it will also play havoc with SFF style motherboards and cases - a x16 slot (and retaining plastic / latch) already reaches almost the end of an ITX board. Plus it's then just connecting a cable from the PSU to the motherboard to then connect that power to the card - why have an intermediary if the original connection was fine? Oh yeah, window cases

To be honest I'm not sure why they pursue using pin + socket connections vs edge style connectors. Even back in the day of old molex (3.5 inch HDD) drives it was a known issue that the male and female pin parts would suffer from spread over time and bending/retightening them back was easy.
ATX / EPS / PCIe connector pins are less susceptible to this but why risk it at all? At heart, SATA, USB, HDMI, DisplayPort, thunderbolt, USB-C, are all using edge connections - some carrying high currents in quite a small form factor - I can't see why a latching edge connector couldn't be used to connect to the PCB of the graphics card..... hell, even something similar to the old original floppy / MFM/RLL connectors but with clips (and without the ribbon cables) using modern design should easily carry that amount of current - essentially no different from the HPCE edge connection and just in a connector form...
It would be easy to do... if they could do it 40+ years ago, they can do it now.
Also, with an edge connector you don't have soldered pin stress/fatigue to worry about on the GPU PCB from repeated connections...

 

[evernessince]

This is based as a percentage. 3.3% of millions of 6-pin/8-pin graphics cards vs 4% of thousands or hundreds of thousands 12VHPWR graphics cards DOES NOT mean that there has been more failures of those within a couple of years vs 15+ years of the existing 6+2pin PCIe power connection existing.

Well first, thank you for taking your time to write a thoughtful reply.

To be precise the data illustrates the number of people who have ever experienced an issue with 6/8 pin as compared to 12VHPWR. 12VHPWR is failing at a higher rate given the total number of failure instances has already exceeded 6/8 pin in the short time it's been available despite 6/8 pin being around for 17 years. As evidenced by your 3.3% of millions vs 4% of hundreds of thousands example, if you have vastly more products on the market using 6 / 8 pin but the total number of people reporting issues is lower, then there is a large discrepancy in the failure rate between the two products.

I've updated this thread title to hopefully more accurately represent what the chart shows.

I'd hazard a guess that more graphics cards that were manufactured with 6+2pin connectors have been scrapped/recycled than 12VHPWR have been manufactured within the last 2 years.

Not exactly sure what scrapped / recycled cards have to do with connector failure rates. If you implying these aren't being included in the figures for 6/8-pin, GN asked people whether they experienced an issue or not, which means people's older 6/8-pin purchases should be included.

It's also based on survey data, not actual recording of hardware failures/returns or the result of some intermediate party that has some reason to keep statistics on such a thing (e.g. an insurance company, etc.), so it's not exactly an infallible data set.

Never implied that it was infallible. This kind of data is the lowest level of scientific evidence but it is often hard to collect more reliable data when you are talking connector issues on people's personal systems. This isn't like servers where if you are lucky and have connections in the industry you can get quite a bit of high quality data (like Level1Techs has on the Intel issues).

Using that same percentage against a sample of 100 of each type would mean that maybe 3 of one type and 4 of the other type would have issues. In any case they would need another 20+ years of both types being manufactured equally before they would even be close of approaching a similar number of overall failures.

I'm not sure it would take that long for failure rates to match, especially if they are as elevated as the data suggests.

I'm not defending the 12VHPWR standard - as soon as I saw it and the power limits designed for it I was like "ok... we'll see how that goes" having already seen failures on the existing 6+2pin system which (as GN itself touched upon) has quite an amount of headroom in its design. To be honest, I was surprised to see so many issues straight out of the gate - I expected it to at least be more of a used / second hand market type of issue.
Also, the simple premise of 'make something better than existed before' has certainly not been the case - if anything 12VHPWR has made the argument for the HPCE + PCIe connector (see link) a much stronger one even if that means backwards compatibility may go out the window or a load of cases with seperate GPU mounts and extension cables need to be replacement parts (unless some cards have an auxiliary port).

I too was under the impression that it would be a used market issue. Particularly given the low mating cycles these connectors are rated for and the propensity of the 3 dimple design's contacts to spread over time. The 3 dimple design of which is still in use as they walked back the requirement to use the newer 4 spring design.

I heard it's not their invention but I got no proof on hand. Anyway, they adopted it so they're also responsible.

And this design is really very poor. I've seen 10 y.o. kids coming up with smarter stuff than that. And they ain't even attending engineering classes.

Wires should be thick, there should be no way you can plug it incorrectly (unless you really try) and there should be 1+ kW abuse capacity in cables rated for 600 W. Bending also should never have been an issue. Enforcing AWG16 so singular old school 6-pins suddenly become suitable for enthusiast segment gaming GPUs would've been a gajillion magnitudes better.

According to the video, it was commissioned by Nvidia and Dell.

 

[Vincero]

Not exactly sure what scrapped / recycled cards have to do with connector failure rates. If you implying these aren't being included in the figures for 6/8-pin, GN asked people whether they experienced an issue or not, which means people's older 6/8-pin purchases should be included.

It's not that it effects the experiences people will had with these cards, just more the point that more of the older style cards have likely been tossed/destroyed than the newer type of cards exist - it's more to highlight the scale that the amount of failures of 12VHPWR cards would be both numerically and statistically low taken as a total.

Never implied that it was infallible. This kind of data is the lowest level of scientific evidence but it is often hard to collect more reliable data when you are talking connector issues on people's personal systems. This isn't like servers where if you are lucky and have connections in the industry you can get quite a bit of high quality data (like Level1Techs has on the Intel issues).

I wasn't being critical of anyone there, just pointing out that this information can't be taken as an absolute - and as you have commented, you have a pretty good grasp of that point but some/many others may not understand survey data is not imperical measurement - otherwise aliens definitely are among us and 9/11 was definitely an inside job.
All GN can say for sure is that, statistically, people who might have some experience with either or both types of connector have a higher experience of some kind of failure of either the card or intermediate connection cable related to the newer 12VHPWR type. The problem with that is, as GN touched upon, if the connector happened to be the Nvidia supplier 3/4xPCIe>1x12VHPWR adapter, part of the problem could easily be a bad connection on the PCIe cabled/connector side having a knock-on effect to the 12VHPWR side... or maybe the PSU maker supplied cable and connection point which may not even be 12VHPWR style... this doesn't help statistics and perception.

I'm not sure it would take that long for failure rates to match, especially if they are as elevated as the data suggests.

Considering you'd have to include graphics cards going back to the Geforce 6000/7000 / Radeon X1xxx era, it would take a very long time if those survey percentages were applied - we are talking the time it's taken for some people to be born and now being able to vote - NOTE: Based on your previous title, going forward then the statement would be true. If Nvidia/AMD/Intel had manufactured enough GPUs in the last 2 years to statistically make a dent on the existing numbers of GPUs there would literally be a flood of products on the market and we'd be back to high end cards in the $500 territory. Literally, RT for everyone...

I too was under the impression that it would be a used market issue. Particularly given the low mating cycles these connectors are rated for and the propensity of the 3 dimple design's contacts to spread over time. The 3 dimple design of which is still in use as they walked back the requirement to use the newer 4 spring design.

Yeah well, we know what these guys think of the used market...

 

[Macro Device]

It was commissioned by Nvidia and Dell

That's irrelevant to me to be brutally frank. I find NV equally guilty in either case.

What matters is that this design is cancer and moronic. Like we both said, forcing PSU manufacturers to only use AWG16 or better cables so 300 W from one 8-pin becomes a prose rather than a luxury will make 99% GPUs only need one cable. Simple and sound.

 

[TheinsanegamerN]

View attachment 366604

This is a photo taken this morning of the CableMod C-Series Pro ModMesh I purchased about a year and a half ago. Only maintenance done is a small drop of DeOxIt D100L on the 12V and ground pins every few months or so.

View attachment 366605

This RTX 4090 is approaching two years of use with zero issues related to power delivery. The card is stock with no BIOS or voltage modifications made.

I don't know what the root cause of the melting connectors is, nor am I going to speculate on it or argue with anyone. I'm just responding to the categoric insistence that it's the fault of CableMod or the fault of the connector, because I'm providing a real-world use case of both of these operating within design specifications.

I did have a melting power connector problem in the past -- it was an ATI All-In-Wonder 9800 Pro where it had one typically used on floppy drives, and it clearly wasn't up to the job.

Funny, I've never had to do maintenance on a plugged in card with 6/8 pin connectors. Never had to clean them, or use deoxit. Never had one melt. Hmmmm....

Poor Steve must be running out of things to create his usual whiny negative content about. Maybe he can find another overheating no-name pre-built at Wal-Mart with poor air flow or go on yet another unhinged rant about Alienware while he's at it. Got tired of his "I'm the smartest guy in the room" schtick years ago.

Getting real "Ow the Edge" energy from this.



"I dont have a problem" is not synonymous with "there is no problem".

That's irrelevant to me to be brutally frank. I find NV equally guilty in either case.

What matters is that this design is cancer and moronic. Like we both said, forcing PSU manufacturers to only use AWG16 or better cables so 300 W from one 8-pin becomes a prose rather than a luxury will make 99% GPUs only need one cable. Simple and sound.

They really should have just made a 16 pin version of the old 8 pin design. That would have solved their problem.

 

[Macro Device]

They really should have just made a 16 pin version of the old 8 pin design. That would have solved their problem.

No. Just ban everything that's worse than AWG16 and disallow 300+ W GPUs into mass market.

I'm talking a problem that never existed anyway. 99.99% users just install their stuff and only rip it off when they upgrade/clean the PC, they don't care how many cords are there.

 

[Totally]

I'm sure it happens, but I've never seen a case of catastrophic failure for either connector that wasn't related to user error in some way.

Yes, blame the user when a large part of design is engineering out user error I.E. before something is stamped as done a bunch of engineers sit in a room solely for the explicit reason of brainstorming, testing all the possible ways a user can f something up, and then revise the design or documentation accordingly.
You might say its all hindsight but this is low-hanging fruit and should have been identified pretty quickly.

 

[thesmokingman]

If it ain't broke, don't fix it.

Words to live by...

 

[DirtyDingusMcgee]

Yes, blame the user when a large part of design is engineering out user error I.E. before something is stamped as done a bunch of engineers sit in a room solely for the explicit reason of brainstorming, testing all the possible ways a user can f something up, and then revise the design or documentation accordingly.
You might say its all hindsight but this is low-hanging fruit and should have been identified pretty quickly.

If the shoe fits.......yeah. If an improperly plugged connector builds heat/resistance and burns up, even if it is not a great design, if is still the fault of the system builder, not the person who designed it. One can never engineer every problem out of a design. In my experience over-engineering just leads to even more problems. I do not love every design choice made by the people that design and build computer parts, but I can't blame them if i screw up.

 

[eidairaman1]

This comes to mind

 

[R-T-B]

of course it's their invention.

The usage case at this wattage and pinout is their invention, mostly.

The connector is factually and provably a molex owned invention.

 

[Vya Domus]

The connector is factually and provably a molex owned invention.

I don't know about that but it's not relevant, it's pretty clear 12VHPWR exists because of Nvidia, they're responsible for it.

 

[Vincero]

I don't know about that but it's not relevant, it's pretty clear 12VHPWR exists because of Nvidia, they're responsible for it.

It would seem so... but I understand the beneficial idea - why have 3 or 4 connectors when they could be replaced with just 1....
As I've said before.... why go with a crappy set of small pins with super tight plastic which can easily break or if bent might compromise it's longevity when there were obviously more robust and simpler options... never mind... I can already see the appeal for a connector that is (almost) guaranteed to grenade itself at some point in terms of new card sales.

 

[A Computer Guy]

All big power GPU's need are doubled up 12VHPWR connectors like GPU's of old. Problem solved for an additional $1 upcharge on a >$600 GPU.

 

[harm9963]

View attachment 366604

This is a photo taken this morning of the CableMod C-Series Pro ModMesh I purchased about a year and a half ago. Only maintenance done is a small drop of DeOxIt D100L on the 12V and ground pins every few months or so.

View attachment 366605

This RTX 4090 is approaching two years of use with zero issues related to power delivery. The card is stock with no BIOS or voltage modifications made.

I don't know what the root cause of the melting connectors is, nor am I going to speculate on it or argue with anyone. I'm just responding to the categoric insistence that it's the fault of CableMod or the fault of the connector, because I'm providing a real-world use case of both of these operating within design specifications.

I did have a melting power connector problem in the past -- it was an ATI All-In-Wonder 9800 Pro where it had one typically used on floppy drives, and it clearly wasn't up to the job.

Poor Steve must be running out of things to create his usual whiny negative content about. Maybe he can find another overheating no-name pre-built at Wal-Mart with poor air flow or go on yet another unhinged rant about Alienware while he's at it. Got tired of his "I'm the smartest guy in the room" schtick years ago.

Do you have a before and after for your GPU 12v rails , and removing the cable too many times , is not recommend ,30 times limit or less ,will be doing a follow up on my cable soon , will do a before and after ,new cable vs old cable , both native 12VHPWR , from same OEM.

​

 

[stahlhart]

Do you have a before and after for your GPU 12v rails , and removing the cable too many times , is not recommend ,30 times limit or less ,will be doing a follow up on my cable soon , will do a before and after ,new cable vs old cable , both native 12VHPWR , from same OEM.

​

Stats at idle.

Stats after 90 seconds of FurMark.

Funny, I've never had to do maintenance on a plugged in card with 6/8 pin connectors. Never had to clean them, or use deoxit. Never had one melt. Hmmmm....

You do you.

 

[harm9963]

View attachment 367133

Stats at idle.

View attachment 367134

Stats after 90 seconds of FurMark.



You do you.

Pulling the cable out too many will affect the 12v rails over time, when new was a solid 12.045 under stress test.

New cable was free from FSP , thanks MR John Chen
https://gpuz.techpowerup.com/24/10/12/c8s.png

 

[eidairaman1]

Pulling the cable out too many will affect the 12v rails over time, when new was a solid 12.045 under stress test.

Yeah because connector ends do expand out, also metal work hardens.