Two NVIDIA graphics with the same name? This is their difference

There is no doubt that in a few weeks we will have the RTX 40 in our hands and we will be able to check its power. However, while we wait for the presentation of the future GeForce, rumors about its specifications and the different models are appearing from the gossip. The last? That we will most likely see two versions of the RTX 4080, one 16 GB and one 12 GB. Is that possible?

There is usually a relationship between the GPU and graphics chip model and the memory configuration used. Since the interfaces that communicate the GPU with the VRAM are located on the outside of its perimeter. So the lower the specs, the lower the interface, and the smaller the amount of memory. Does it make sense to see disparate configurations? Well yes, but they are usually sold as different models and it is usually done by removing memory chips from the card. Are we really going to see different configurations for the RTX 4080?

Two NVIDIA graphics with the same name

RTX 4080 16GB and RTX 4080 12GB?

The amount of memory also indicates the number of chips and thus the bandwidth, since in the case of GDDR6 and GDDR6X each of them has a 32-bit bus. So for a 12GB RTX 4080 card at 2GB per chip, this is a 192-bit bus versus the 256-bit the 16GB model would have. That is, a 33% bandwidth advantage, as long as the memory speed is the same. As if that were not enough, it is likely that the differences do not end there, since we would also be talking about two different PCBs and, therefore, two different graphics card models. The first made up of 10 layers and the second of 12. Which is usually done when we have two different chips.

Render GPU flotando NVIDIA

Let’s not forget that a few months ago, through the hack made to NVIDIA, the almost final specifications of its RTX 40 were leaked, specifically of the different chips. In them we could see, apart from the already known 144-core AD102, also models with lower specifications. Among those that stood out were the AD103 with a 256-bit bus and 84 cores in total, and the AD104 with a 192-bit interface and 60 cores. If the rumor is true, then this would mean that NVIDIA has decided to rename the potential RTX 4070 as RTX 4080 12 GB. Something that is unprecedented. In any case, it would not be the first time that NVIDIA has proposed specific models and then discarded or launched them later in time.

First image of the graphics card?

During these days, the possible first image of the RTX 4080 has also been leaked, whose design is very reminiscent of that of the current RTX 30 . Which should not surprise us since everything indicates that only the cards that use the AD102 will need the high consumption provided by the new 12+4-pin connector. So this model should be between 325 W and 375 W. In any case, it is expected to outperform the entire range of current cards, including the RTX 3090 Ti.

Posible Imagen RTX 4080

In any case it could be a 3D print or a modified version of the RTX 3080 made with a 3D printer. The fact that it is covered with a plastic to make it difficult to see the type of plastic that covers the card seems suspicious to us. In any case, it should not be long before we can finally see them and see how NVIDIA intends to continue to dominate the market for another generation.