What about the hype of these new graphics cards? Lots of leaked data, lots of performance promises and not a single test yet, pure speculation keeping gamers up at night. Well, if you’re one of those people who isn’t going to spend an average of $1300 on a GPU and already assumes you have a TOP PSU that supports these beasts, get used to the idea that changing the power supply is almost mandatory for all models.
NVIDIA will boost the consumption of its RTX 40 graphics cards
There is not too much secrecy here except on the values as such, since thanks to a prestigious leaker we know today what we will have to stick to in just a few months. That the top end consumes an outrage is even to be expected if you’ve followed us chronologically over the course of this year, but what few might hope is that performance will match consumption extrapolating the data between the RTX 30 and the RTX 40.
April Fool,
**90 PG139-SKU310, AD102-300, 24G 24Gbps, 600W;
**80 PG139-SKU330, AD102-200, 20G 24Gbps, 500W;
**70 PG141-SKU310, AD104-400, 12G 24Gbps, 400W.
🫣🫣🫣If I’m wrong, please don’t kick my ass. 🤣🤣🤣
– kopite7kimi (@kopite7kimi) April 1, 2022
Assuming today is April Fool’s Day, I hope that was a joke, but no, nothing more. What was revealed by Kopite7kimi shows three classes of typical NVIDIA graphics cards called 90, 80 and 70 where there are some pretty interesting details to comment on, but let’s start with the power consumption of each GPU line.
The presumed RTX-4090 they will go to the mass of 600 wattswhich would imply a power supply of 1200 watts or more if we are to maintain efficiency at maximum levels above the 90%nonsense that we hope lives up to consistent incremental performance.
the RTX4080 lower the bar a foot and will need a 1000 watt power supply or more as they will consume 500 watts per unit, while RTX4070 increase the consumption of the mid-range to 400 wattsnonsense for the gamer who wants to scratch his pocket because he will have to get a quality power supply of more than 800 watts.
Interesting details to note
Considering the consumption and what it will mean for users who wish to checkout (we will see the prices as such), there are some rather striking things to comment on. Starting with the two SKUs and PCBs that will be needed, which are disparate and more than curious.
The RTX 4090 and RTX 4080 share a PCB, but not a SKU, which they do between said RTX 4090 and RTX 4070.
The chips are also surprising, since the RTX 4070 will not initially include the AD103 As expected. Apparently, NVIDIA would reserve it for a new SKU which might be a response to AMD if it finally comes into full force in the high-end competition.
Finally, the capacity of the VRAM varies from 24 GB, one 20 GB and finally to 12 GBthen book it AD103 with 16 GB It makes a lot of sense. In any case, NVIDIA will entrust everything to the 24Gbps GDDR6X VRAM in all the models, it shows that they are at the limit of the bottleneck in this section. Given this, the main thing remains to know, prices and performance, but it is still too early for that.
