So far, Nvidia has launched Volta, Ampere, Hopper and other architectures for high-performance computing and AI training, and launched high-end Gpus such as V100, A100 and H100 on this basis, and the double-precision floating-point computing power for vectors has gone from 7.8 TFLOPS to 30 TFLOPS.
The establishment of the market advantage is attributed to the balance and improvement of the ecology of Nvidia GPU product capabilities.