Anticipation is growing around the next generation of high-end graphics cards from Nvidia, the GeForce RTX 50 series. Official presentations are normally expected at the next CES scheduled for January 6 in Las Vegas.
Indeed Nvidia has scheduled a major press conference during which Jensen Huang, the CEO, could unveil this highly anticipated series. It is not certain that the announcement will be followed by immediate marketing, however it will provide valuable details around the Blackwell graphics architecture, succeeding Ada Lovelace who gave life to the GeForce RTX 40 series in 2022.
GeForce RTX 5090, la vitrine
At the top of the range, the GeForce RTX 5090 is supposed to reveal itself through muscular specifications. If the rumors are correct, the beauty will be equipped with 32 GB of GDDR7 operated through a 512-bit bus to ensure a bandwidth of 1,792 GB/s, all accompanied by a GPU with 21,760 CUDA cores.
The GeForce RTX 5080 should benefit from 16 GB of GDDR7 and a 256-bit bus in order to achieve a memory bandwidth of 1,024 GB/s. To this would be added a GPU with 10,752 CUDA cores. The GeForce RTX 5070 should rely on 12 GB of GDDR7 operated via a 192-bit bus.
On the other hand, if the increase in performance promises to be significant, it is likely that the energy demand will follow the same trend to the point that the GeForce RTX 5090 will be capable of consuming 600W, compared to 450W for the current GeForce RTX 4090 and 400 Watts for GeForce RTX 5080.
Finally, as for prices, they give rise to discussions. The latest reports on the subject put forward unconfirmed but significant ranges:
- the RTX 5090 could be between $2,000 and $2,500,
- the 5080 between 1,200 and 1,500 dollars,
- and the 5070 between 600 and 700 dollars.
This prospect is not encouraging because it is reminiscent of the upward price trend observed in recent years in the graphics card sector…
