The video card's sales of GeForce RTX 408016GB are due to start next month. However, China's Chipwell Forum has already produced data on accelerator performance. It has been tested in synthetic tests AIDA64, 3DMarc Fire Strike Ultra, 3DMarc Time Spy Extrame and 3DMarc Port Royal, as well as in a 3DMarc test for Intel XesS scale technology. Finally, the new one has passed one game test.
According to AIDA64, the graphic processor of the video map operates at a frequency of 2505 MHz and has 76 multi-processors. According to the benchmarking, the video map provides a level of output of 53620 Gflops in single accuracy operations, just above the level promised by the NVIDIA itself.
The map performance is also impressive in synthetic 3DMark tests. With 17465 and 13997 in 3DMarc FireStrike Extrame and TimeSpy Extrame, the GeForce RTX 408016GB booster was the fastest among all the video cards involved.
Insiders also shared the performance data of GeForce RTX 408016GB in Intel Xess and Port Royal tests. In the first case, it was not much to compare with because the test had just come out yesterday, and Port Royal was faster than all the video maps of the previous generation, and had significantly outrun even the flagship GeForce RTX 3090 Ti.
The map was also tested in Shadow of the Tomb Raider at 4K resolution. According to the game test, the card showed an average FPS between 87 and 115, with a maximum FPS of 226.
NVIDIA introduced two versions of GeForce RTX 4080. The 16 Gbyte memory model is based on the AD103 graphic processor, which contains 9728 CUDA cores. The 12 Gbyte memory model uses the AD104 graphic processor with 7,680 CUDA nuclei. Both versions are due for sale next month. However, NVIDIA has not yet specified the exact date of their release.