Nvidia's top-of-the-line next-gen graphics card can end up with over 80 TFLOPS of performance. According to the source, the GPU AD102 of the Ada Lovelace generation, which will serve as the heart of the GeForce RTX 4090 adapter, will operate at a frequency of 2.2 GHz. Probably referring to the Boost frequency. For comparison, the RTX 3090 is less than 1.7 GHz. As a result, with the agreed frequency and the presence of 18,432 CUDA cores, which various sources have already mentioned more than once, the performance is about 81 TFLOPS. The RTX 3090 has 35.58 TFLOPS.
The source also talks about the 5 nm process technology and the 384-bit bus adjacent to the GDDR6X memory. Apart from the process technology, there are no differences from the current flagship Nvidia. Earlier, we recall, sources talked about high power consumption of new AMD and Nvidia cards, attributing the RTX 4090 model a TDP value of more than 400 watts. While these are all rumors or leaks, the sources are often reliable enough. In this case, the information was shared by an insider who more than once published data on AMD and Nvidia video cards, which were later confirmed.
0 Comments