NVIDIA GeForce RTX 4090 GPUs Burning Their 12VHPWR Power Connectors
27 October 2022

Why is it that every single time NVIDIA makes headlines, more often than not for barbaric bullshit or blatant kikery that it never just ends there, it’s never just one thing going on with NVIDIA at any one time but more rather time and time again when they slip up they instead tumble down the stairs.

The NVIDIA corporation over the past few months have seemingly stopped trying to hide their anti-consumer tactics, but of course the general consumer is none the wiser to their bullshit, considering how the most kiked corporation in this particular field still maintains a market share of over 80 percent.

The GeForce RTX 4090 recently released a few weeks ago, it offers thereabouts 60% greater rasterized performance over the RTX 3090 Ti which all things considered is possibly the greatest leap in generational GPU uplifts, however there’s plenty wrong with ADA Lovelace.

The fact that the RTX 4090 has an MSRP of $1600, is a severely cutdown AD102 core which leaves NVIDIA wide open to a superior RTX 4090 Ti down the line, doesn’t offer DisplayPort 2.X functionality, instead maintaining DP 1.4 connectivity as well as recycling the overengineered PCB design of the RTX 3090 Ti.

ADA Lovelace introduces cancerous DLSS 3.0, whereas the peasant tech of yesterday merely “upscaled” blurry TAA riddled trash this latest rendition actually generates frames through interpolation.

The results are hilarious, nevermind the fact that DLSS standalone increases input latency by a significant degree, with frame interpolation essentially generating its own frames in queue it undoubtedly increases latency furthermore, which is why NVIDIA REFLEX is enabled by default with DLSS 3.0

As the frames generated with DLSS 3.0 are an interpolation of what the next game “should be”, rather than an actual in-engine rendered frame, NVIDIA themselves have claimed that the overall latency is “similar to native rendering” in some times, it’s actually possible for some games to offer lower latency, but much like how DLSS from the very beginning has incorrectly been touted as being “better than native”, majority of the cases with DLSS 3.0 will show latency being worse than native rendering.

In fact the input delay could in theory be doubled, if you doubled your framerate thanks to DLSS 3.0 but every other frame in the queue, since it’s interpolated, cannot actually display changes to an input between the source and interpolated frames, you may have “doubled” your framerate but overall speaking you’re still running at the same framerate in terms of latency.

Discrepancies can occur when testing of course, because if the moment you click a button during a rendered frame you’d have to wait two frames minimum for it to actually display in game with DLSS 3.0

The cancer that is DLSS 3.0 aside, not only has NVIDIA provided a massive performance uplift with the RTX 4090 but with ADA Lovelace comes a serious increase in the overall power consumption, leading to several units of the RTX 4090 burning the 16-pin power connector.

The 16-pin power connector found on several in-house RTX 3000 series designs, otherwise known as the 12VHPWR connector is Intel’s newest staple with the ATX 3.0 format, NVIDIA first adopted this new standard for the RTX 3090 Ti and have followed through with the 12VHPWR connector for the RTX 4090 GPU.

The RTX 4090 is a hungry behemoth, with most aftermarket cards coming in at around 4 slots in terms of thickness, with more than adequate PCB designs being offered across all units, the RTX 3090 Ti never had any such issues but it seems the ~450W RTX 4090 are generally burning up.

From the beginning the 12VHPWR connector was rubbish, while nobody generally goes about plugging in and unplugging their power connectors for the heck of it, you’re generally only able to disconnect and reconnect a 16-pin power connector around that of thirty times maximum, which shows how fragile they really are.

Generally speaking, offering a 3x 8-pin power requirement allows a maximum of 450W to be drawn, with those connectors in total providing nine 12V pins inside, whereas the 12VHPWR connector can reach a theoretical limit of 600W from just six 12V pins.

Safety has generally been thrown out the window here as NVIDIA generally doesn’t care anymore, because of course if NVIDIA opted to avoid this new ATX 3.0 Gen 5 standard they’d have to increase the PCB size of the RTX 4090 by just a little bit more and obviously, we cannot have that.

Several instances of aftermarket GPUs from ASUS, specifically the TUF Gaming and Gigabyte are literally going up in smoke and burning their power connectors.

While we don’t exactly know if this is generally a problem that will simply happen to just about anyone, people have been chiming in that the main cause of the destruction has stemmed from 12VHPWR cables being “bent” inappropriately too close to the terminal itself.

Which is damn near impossible considering the sheer size and girth of the RTX 4090 graphics card, it’s damn near impossible to fit the connector onto it without bending the cable in the first place, all because NVIDIA wished to be more compact in pushing more power through thinner pins, saving merely shekels in additional plastic and circuitry.

NVIDIA would have been better off fitting the RTX 4090 with an AC power connector, allowing consumers to plug their expensive and hungry graphics card directly to a wall outlet, god knows just how much worse this bullshit is going to be once NVIDIA unload the RTX 4090 Ti which will undoubtedly pull near 600 watts in total.

A potential solution perhaps?

Imagine spending over $1600 just for NVIDIA to shit into your open mouth with a literal fire hazard you cannot afford to power on during these times, because the damn thing chugs wattage like a frat kid does alcohol but you cannot actually afford to leave it powered on unattended just in case yours decides that it wishes to melt its own power connector in which case if not caught quick enough will undoubtedly lead to a house fire situation.

blog comments powered by Disqus