Nvidia GeForce RTX 5090: Rumors About The Performance of New Graphics Card

Although the new Nvidia graphics cards in the RTX 5000 family are not expected until 2025 at the earliest, rumors about the possible GeForce RTX 5090 are currently abuzz. Insiders speak of a drastic increase in computing units, memory, clock speed, and performance.

More memory, clock speed, and performance

In the infamous Chiphell forum, the leaker “Panzerlied” is currently talking about the possible specifications of the upcoming Nvidia GeForce RTX 5090 in direct comparison to the current flagship, the RTX 4090. The often well-informed insider is leaning way out of the window when you consider that Nvidia’s “Blackwell” generation will not appear until 2025.

If you trust his forecasts, the clock speed of the RTX 5090 could increase by 15 percent compared to the RTX 4090. An increase of more than 50 percent is also planned for the computing units and memory bandwidth, while the L2 cache could be increased by 78 percent. In combination, a general performance increase of 70 percent is currently expected.

If the rumors are true

The colleagues at Videocardz added the percentages to the current technical data of the RTX 4090 in order to create a data sheet for the Nvidia GeForce RTX 5090 based on the early leaks. Accordingly, the graphics card with GB202 GPU would have 24,000 CUDA cores, a boost clock of 2.9 GHz, and up to 32 GB of GDDR7 memory.

Furthermore, the potential RTX 5090 could work with a 512-bit memory interface and a throughput of up to 32 GB/s. The storage bandwidth would also be over 1.5 TB/s. The L2 cache is up to 128 MB. The associated Total Graphics Power (TGP), which is specified as 450 watts for the current RTX 4090, remains unclear.

Since we are more than a year away from the market launch of possible Nvidia graphics cards with “Blackwell” architecture, the rumors should be viewed with a certain degree of skepticism. More tangible leaks are likely to occur over the next year.