NVIDIA is currently assembling its AD102 chip with Ada Lovelace architecture for its future RTX 40. A chip that until relatively recently we related to both the RTX 4080 and the RTX 4090. However, the latest information that has been leaked has told us confirmed that NVIDIA will return to the tradition of previous generations of using different chips for each of the components in the range.
From the leak last February, the key question was none other than where the AD103 chip would end up, well, now we know the answer to where this variation of the future GeForce Lovelace will end up.
Why will there be such a difference in performance between the RTX 4080 and RTX 4090?
Through a tweet from Kopite7kimi, who is one of the most reputable information leakers when it comes to NVIDIA, we have been able to learn new information about the way in which the different graphics chips or GPUs will be distributed under Lovelace architecture in their future RTX 40. Information that contradicts that given a few months ago regarding the next generation of graphics cards from the Jensen Huang company.
In the current RTX 30 we have been able to see how the top of the range chip, the GA102, which has been used in the RTX 3080 (Ti) and 3090. Instead, it seems that in the case of the chip AD102 would only be used on the RTX 4090so the AD103 would be seen on the RTX 4080. What does this mean? Well, there is going to be a gap in performance between the most powerful model of the next generation and the second most powerful model. The reason? If Kopite’s information is confirmed, then the RTX 4090 will have 144 SMs and a 384-bit bus, while the RTC 3080 will have 84 SMs and a 256-bit bus.
If we talk about energy consumption, it is clear that the RTX 4090 will move around 600 W, but in the case of the RTX 4080 there are more doubts. Let’s not forget that the GA102 GPU has moved in consumption of 375 W and 450 W in the current range of graphics cardsyes This indicates that we could, on the one hand, a RTX 4080 with a TGP of 375 W and on the other one RTX 4080 Ti with a TGP around 450 W. Although we would not expect such a Ti model among those that will be launched initially.
The price will also be affected
The difference in power between the RTX 4080 and the RTX 4090 will also affect its price. Given that this will be much larger than the one between the RTX 3080 and RTX 3090 at launch, the bottom line will be that having NVIDIA’s next top of the line won’t be cheap to come by in terms of money. If we take into account that it will share space with the RTX 3090 Ti that has an MSRP of 2000 dollars, is it possible that the RTX 4090 will move above that figure? What is clear to us is that the RTX 4080 could be cheaper than expected.
On the other hand, the situation of the RTX 4080 based on the AD103 chip is different, without counting optimizations that we still do not know, its power seems to be close to the RTX 3090 Ti, but consuming much less and at a lower cost. The advantage over the current top of the range will be to have more memory. Our prediction? The most powerful of NVIDIA cards today will disappear from the market in a few months, what’s more, we believe that internally NVIDIA has already marked the end of the production of all graphics cards based on the GA102 chip and that they will soon be replaced by the new range.
The NVIDIA RTX 40 will be more expensive than the RTX 30, mostly due to the increased cost of the new manufacturing node. A price increase that will be offset by greater power. We do not consider the RTX 4090 to cost more than 2000 dollars, but much more than the RTX 3090 at its launch date.