With the RTX 4000 family, Nvidia will take the competition in the video card market one step further. New details of the flagship RTX 4090 have emerged.
According to rumors, AMD will offer 4 times better single-core performance than the previous generation in RDNA 3 architecture. Wanting to attract more customers in the graphics card market, AMD will challenge Nvidia with its flagship model. It is claimed that Nvidia will come up with a similar upgrade for users.
Nvidia RTX 4090’s FP32 will be 100 TFLOPS
The FP32 performance of the RTX 3090 Ti is rated at 40 TFLOPS, while the RTX 3090 is rated at 35.58 TFLOPS. It is stated that the AD102 GPU, powered by the Ada Lovelace architecture , will have FP32 performance of more than 2x.
The AD102 GPU will have 18,431 CUDA cores to achieve 100 TFLOPS. The new model is also expected to be offered to users at 2.7GHz. It is stated that the Navi 31 GPU developed by AMD will be introduced with a speed of 3.0GHz.
The AD102 GPU will get its power from cores produced by TSMC with 5nm technology. It is worth remembering that the previous series used cores produced by Samsung with 8nm technology.
Nvidia used 84 Streaming Multiprocessors (SM) in the RTX 3090 Ti. It has leaked that the company will show up with 144 SMs on the RTX 4090.