NV 12VHPWR issues revisited

Page 27 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Jan Olšan

Senior member
Jan 12, 2017
506
981
136
Memes aside, could this be the start of issues at Nvidia in general?

I mean, they have almost 30.000 employees, is there really not a single person that looked at the 4090 cable issues and analysed the new cards? It's hard to believe ...
I believe the engineers know well this is bad and objected. This issue is likely down to the looks & design language "johny ive" types having the last word and overriding engineer's safety&responsibility concerns and well, knowledge

The FE is a 2 slot design. Thats the problem. No room for power connector safety features.

I'm pretty sure that's not true. just making the card 1 cm longer would for example make space for a second vertical connector, but spreading the heat between two connectors so that it's all safer is a huge nono for Nvidia's design guys apparently. The goals the card's design follow are just not aligned with such lowly things.
 
Last edited:

NTMBK

Lifer
Nov 14, 2011
10,397
5,620
136


This is how our GPUs should connect. Nvidia shoves 700W through those mezzanine connectors with absolutely no issues.

The modern ATX PC is a pile of hacks on top of hacks on top of hacks, on top of a chassis design from the 1980s. The AT motherboard was originally designed in 1984 for this bad boy:



That's where the original layout of "small board mounted CPU and a bunch of expansion slots on riser cards" came from. When your CPU was an 80286 that used 3.3W, just dumping the heat straight into the chassis made total sense! Not like some idiot is going to shove a 250W CPU in here, right? And we need all those slots for network cards, sound cards, modems, passively cooled single slot graphics boards, etc. Those boards are all low power consumption, no need to put ventilation around them, doesn't matter they're cramped together, passive cooling from the warm air circulating around them will suffice.

The board got shuffled around in the 90s with the ATX standard, but the same basic layout remains. It's entirely unsuitable for 500W GPUs on riser cards, and 250W CPUs on the board. We've tried to work around it with insanely large GPU coolers, water cooling loops just to move the CPU heat to the outside of the case instead of dumping it inside, crappy power cables dangling over the board to deliver enough power to these monster daughterboards looming over the rest of the system. Is anyone surprised we ended up here?



We need an entirely new architecture actually designed for the high power CPUs and GPUs of today, which will let us build smaller, simpler designs with sane power delivery and cooling.
 

Heartbreaker

Diamond Member
Apr 3, 2006
4,742
6,247
136
The modern ATX PC is a pile of hacks on top of hacks on top of hacks, on top of a chassis design from the 1980s. The AT motherboard was originally designed in 1984 for this bad boy:
...
We need an entirely new architecture actually designed for the high power CPUs and GPUs of today, which will let us build smaller, simpler designs with sane power delivery and cooling.

100% This. But changing from an entrenched standard is massively difficult.
 
Reactions: beginner99

MrTeal

Diamond Member
Dec 7, 2003
3,869
2,524
136
Absolutely, and a lot of people have been saying so for quite awhile. Besides the obvious electrical advantages of mounting a mezzanine connector to the motherboard, you move the board out of the way and would be able to use a flow-through tower cooler.

Plus a whole GPU/CPU/VRM monoblock would look awesome.
 

DrMrLordX

Lifer
Apr 27, 2000
22,475
12,331
136
Idk what's worse, the meme or sending those images as PSD files ....

PTSD from PSD?

There is nothing sensible about that. How about do what adult Jimmy Neutron said, and go back to the tried and true method. If you need 4 cables plugged in? so be it. If buyers need to buy a new PSU designed for the new cards? so be it. If you are already spending $2500 or more on a card, you can pony up a few $$$ for a new PSU too.

Apparently NV wanted a specific PCB layout for the 5090 FE (among other cards) and couldn't manage a larger PCB to achieve their objective. So they were stuck with a single power connector and no shunt resistors.

User error was a big part of the problem back when GN duplicated melting with partial insertion, and then you could see that most of the melted connectors had marks indicating connectors weren't properly installed.
A modern connector that tries to carry 450-600W on a single cable with 14% margin needs failsafes. A partially-inserted cable should have either signaled to the card to a). reduce clocks/power or b). not function at all. Clearly both 12VHPWR and 12V-2x6 don't require these features, leaving it up to the implementation, and even then, the best implementations (such as ASUS's 5090s) have their flaws.

There's no way you should be able to cram 50a down one wire/pin of 12VHPWR/12V-2x6 by cutting some wires.
 
Last edited:

Thunder 57

Diamond Member
Aug 19, 2007
3,403
5,625
136


This is how our GPUs should connect. Nvidia shoves 700W through those mezzanine connectors with absolutely no issues.

The modern ATX PC is a pile of hacks on top of hacks on top of hacks, on top of a chassis design from the 1980s. The AT motherboard was originally designed in 1984 for this bad boy:



That's where the original layout of "small board mounted CPU and a bunch of expansion slots on riser cards" came from. When your CPU was an 80286 that used 3.3W, just dumping the heat straight into the chassis made total sense! Not like some idiot is going to shove a 250W CPU in here, right? And we need all those slots for network cards, sound cards, modems, passively cooled single slot graphics boards, etc. Those boards are all low power consumption, no need to put ventilation around them, doesn't matter they're cramped together, passive cooling from the warm air circulating around them will suffice.

The board got shuffled around in the 90s with the ATX standard, but the same basic layout remains. It's entirely unsuitable for 500W GPUs on riser cards, and 250W CPUs on the board. We've tried to work around it with insanely large GPU coolers, water cooling loops just to move the CPU heat to the outside of the case instead of dumping it inside, crappy power cables dangling over the board to deliver enough power to these monster daughterboards looming over the rest of the system. Is anyone surprised we ended up here?



We need an entirely new architecture actually designed for the high power CPUs and GPUs of today, which will let us build smaller, simpler designs with sane power delivery and cooling.

Isn't that what BTX was supposed to be?
 

MrTeal

Diamond Member
Dec 7, 2003
3,869
2,524
136
Isn't that what BTX was supposed to be?
BTX was designed to help improve CPU cooling in the P4 days. It moved the CPU towards the front of the case so you could duct fresh intake air right to it.

It didn't to much of anything for PCI cards, it was a response to the crazy days when a Prescott CPU could draw all the way up to 100W!
 
Reactions: KompuKare

DrMrLordX

Lifer
Apr 27, 2000
22,475
12,331
136
Yup, BTX featured side case fans that were supposed to feed the HSF. Wacky design actually, given how small the side case fans were.
 

amenx

Diamond Member
Dec 17, 2004
4,219
2,551
136
I dont get it. Why would Nvidia not implement something similar and at low cost? Why allow a worse alternative dealing with RMAs and damaged rep?
 
Jul 27, 2020
23,127
16,297
146
I dont get it. Why would Nvidia not implement something similar and at low cost? Why allow a worse alternative dealing with RMAs and damaged rep?
I think Nvidia refuses to take responsibility for RMAs. They probably have some sort of agreement where the AIB runs a really crazy stress test and if that is passed and certified, the AIB assumes all responsibility for the failure of the GPU/VRAM kit since they ran the strenuous test and there's maybe 1 in 1000 chance of that kit failing after passing the test, odds that the AIB just accepts as cost of doing business with Nvidia. I think this is why EVGA bailed because they must have recognized the inherent problems in the 4090 design and didn't want to deal with unprecedented volume of RMAs.
 

amenx

Diamond Member
Dec 17, 2004
4,219
2,551
136
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |