Sneaky Nvidia

T4rd said:
Sneaky indeed.. pretty much invalidates and makes all previous 4090s obsolete. And how will current 4090 owners know which card to order for the "upgrade"?! /s
Click to expand...
Yeah, that's a nice and quiet admission that their original power connector actually was the piece of shit that most of us knew it was. They did something like this back in the 2080Ti era where they released a shitload of defective silicon and lied about their failure rates. I think it was like a 20+% Failure rate instead of the 1-2% they claimed it was.

I like their graphics cards but you can't trust anything they put out until at least a year after it's been released and vetted to be stable. AMD may not be as good and they have their own issues with stuff like the 7900XTX launch but they at least honored replacing the cards within short order.
 
Westwood said:
Watch your homeowner's insurance go up. :D
Click to expand...
TBH - If my house went up in flames because of a 4090 burning up, I'd absolutely get with my insurance company to sue the shit out of Nvidia. Them changing this connector is basically an admission of guilt.
 
Pretty sure that the last 4090Fe's that were in stock with the original version of the 12vhpwr connector were sold by BestBuy in December 2022. None have been stocked for months, up until the end of June. Those cards have visibly shorter sense pins. Whether or not the connector fully matches the new specs is unknown, but it probably does.

The shorter sense pins is known so far, only on newly shipping Nvidia branded cards. No idea what the board partners are doing, but I would bet they are on track to make the same swap for the part at any time.
 
GoldenTiger said:
Just because something is being improved doesn't mean the old one wasn't fine but for user error.
Click to expand...
The connector was a poor design because it so easily allowed user error. That's the issue. You don't design a consumer grade part with a connector like that. I have a 4080, and that connector blows. Further, it has very limited cycles compared to a standard PCI-E 8-Pin.
 
Mchart said:
The connector was a poor design because it so easily allowed user error. That's the issue. You don't design a consumer grade part with a connector like that. I have a 4080, and that connector blows. Further, it has very limited cycles compared to a standard PCI-E 8-Pin.
Click to expand...
How often do you reseat your card? I usually do it a grand total of once, to move it to another system as a hand me down. Unless you're a hardware reviewer that's a silly thing to get hung up on :).
 
My only gripe with the original connector was the rats nest of cabling that didn't allow the side panel to close.
 
noxqzs said:
My only gripe with the original connector was the rats nest of cabling that didn't allow the side panel to close.
Click to expand...
Which is only an issue using the adapter since it requires four 8-pin cables.
 
Armenius said:
Which is only an issue using the adapter since it requires four 8-pin cables.
Click to expand...
I concur. I decided to take matters into my own hands. Been running like this for the better part of the year now with zero issues and the thing is rendering daily.
 

Attachments

  • PXL_20221110_202408360.jpg
    PXL_20221110_202408360.jpg
    427.3 KB · Views: 0
Mchart said:
The connector was a poor design because it so easily allowed user error. That's the issue. You don't design a consumer grade part with a connector like that. I have a 4080, and that connector blows. Further, it has very limited cycles compared to a standard PCI-E 8-Pin.
Click to expand...
Yeah, we should totally blame a company because users cannot seat a connector properly! /s

No one ever takes accountability anymore. You can always spot the American with the instant "I'll sue!". Grow up.
 
Armenius said:
When do companies ever announce such minor revisions to the public?
Click to expand...
Almost never. Out of all the years I've been in the hobby and professionally reviewing and the two plus decades I've serviced computer hardware I've seen many revisions most people have never heard of. The only exception to the rule was with ABIT and GIGABYTE who announced major board revisions like the BX6 Rev 2.0 or anytime GIGABYTE did v1.1's or 2.0's of some of their boards. Though the latter had almost no fanfare and ABIT treated such boards as separate and newer models. However, Creative Labs NEVER said shit about them removing PCB features and redesigning cards with cheaper components over time. Neither did Adaptec or 3Com who all did the same thing. Intel never said shit about changing the south bridge cooler on the D5400XS, or anything like that. Some ASUS boards even had PCB revisions for minor things.

I've also seen video card PCB's get changed. OEM's like Dell and HP do this sort of thing so often it would completely blow the minds of people who have never serviced systems like those. There are often five part numbers for one thing in an HP system and I'm not counting the option and assembly numbers.
T4rd said:
Sneaky indeed.. pretty much invalidates and makes all previous 4090s obsolete. And how will current 4090 owners know which card to order for the "upgrade"?! /s
Click to expand...
No, it doesn't. It makes the newer models more desirable but it doesn't make the older ones obsolete. People will also figure out which cards to buy though more casual consumers will still be breaking the old ones until they are gone.
Mchart said:
TBH - If my house went up in flames because of a 4090 burning up, I'd absolutely get with my insurance company to sue the shit out of Nvidia. Them changing this connector is basically an admission of guilt.
Click to expand...
Damn right. I had this thought myself. I'm not normally someone who favors the litigious nature of American society but there are cases where I think such lawsuits have merit.
GoldenTiger said:
Just because something is being improved doesn't mean the old one wasn't fine but for user error.
Click to expand...
This is true.
Mchart said:
The connector was a poor design because it so easily allowed user error. That's the issue. You don't design a consumer grade part with a connector like that. I have a 4080, and that connector blows. Further, it has very limited cycles compared to a standard PCI-E 8-Pin.
Click to expand...
I can see both sides of this. However, a design that allows for so many failures and problems is a bad design. That said, customers will figure out how to break stuff doing any DIY project.
 
The whole shenanigans with this connector is why I live an N-1 lifestyle. I'll let other folks be guinea pigs. I rarely have the latest and greatest of anything. Early adopters can burn their houses down, I'm good.
 
DooKey said:
Let me know when a house burns down from this user problem.
Click to expand...
I love when both people in an argument are disingenuous. Calling it a user error when it's bad product design is absurd, so is claiming anyone burnt their house down over it. It was a bad design and they're rectifying it, if they thought it was just user error and not a piss poor design that led to error they wouldn't be rectifying the problem.
 
GoldenTiger said:
How often do you reseat your card? I usually do it a grand total of once, to move it to another system as a hand me down. Unless you're a hardware reviewer that's a silly thing to get hung up on :).
Click to expand...
As far as I know the plug in cycles are the same rating as the older 8-pin - something like 30 cycles?

Anyway I think there's blame to go on both sides here. Definitely user error, but the design could be more robust in its securing mechanism to help mitigate that. Especially considering the fact it is higher power vs the older standard. So why should the securing mechanism be flimsier? I don't think there is an either or scenario here like I've seen argued to death where the one side argue all day long how it can't possibly be a design issue and the other side says it can't possible be a user error. But the fact they revised it at least tells me that they recognized the design could be better to help with user error...which seems more prevalent on this connector...due to design? See how this all fits in together.

I guess what I am saying is user error and a design that could be better can both be true at the same time. I don't understand the die hard "it can only be one of these options" arguments.
 
You must log in or register to reply here.
Back
Top