The appearance of the RTX 40 series graphics card is becoming clearer.

The Electronic Times revealed in its latest report$NVIDIA Corp (NVDA.US) $The next generation GPU includes two major architectures, Hopper and Ada Lovelace, both of which are based on Taiwan Semiconductor Manufacturing Co Ltd's 5nm process.

Among them, Hopper is mainly for data centers, serving AI, cloud computing, meta-universe and other scenarios, with multiple new MCM designs, and will use Taiwan Semiconductor Manufacturing Co Ltd's CoWos fan-out package, which is expected to flow soon.

Ada Lovelace is the game graphics card GPU, corresponding to the RTX 40 series.

WCCFTECH even quoted a reliable source as saying that the RTX 40 series confirmed Taiwan Semiconductor Manufacturing Co Ltd 5nm.

It is reported that AD102 GPU is expected to integrate 18432 CUDA cores, acceleration frequency 2.3~2.5GHz, single-precision floating-point 85~92TFLOPSThat is to say, the maximum energy can reach 2.5 times that of RTX 3090 (36TFLOPS).The more distant RTX 2080 Ti is only 13T.

According to the current progress, the RTX 40 series graphics card will be released as scheduled in 2022, most likely in the third quarter.