Both the Ada Lovelace (Nvidia) and RDNA 3 (AMD) graphics architectures will take advantage of the 5nm node technology manufactured by TSMC and possibly Samsung.
According to the well-known source kopite7kimi, both Nvidia and AMD are currently working on their next-generation graphics cards. On Nvidia’s side, they are working on the Lovelace architecture. On AMD’s side, the RDNA-3 architecture is being used.
Nvidia Ada Lovelace and AMD RDNA 3 would be 2.5x faster than RTX 30
According to information from kopite7kimi, the upcoming graphics cards from Nvidia and AMD should have 220% and 250% performance compared to the RTX 30 series.
It is estimated that the high-end parts could have a performance of 100 TFLOPs (FP32/SP), unlike the GeForce RTX 3090, which has a performance of 35.58 TFLOPS FP32.
The RTX 40 (Ada) series would be more than twice as fast as the RTX 3080 and RTX 3090 GPUs, while the Navi 31 RDNA 3 chip could be slightly faster.
According to Kimi’s estimates, the high-end RTX 40 series GPU (Ada) will be more than twice as fast as the GeForce RTX 3080/3090, while AMD’s flagship Navi 31 could even be faster than Ada Lovelace. Furthermore, Nvidia’s chip could be the first to feature an MCM design.
If AMD manages to make its next generation of graphics cards superior to Nvidia, it would be a huge success for the red company, which has always lagged behind Nvidia in terms of performance in recent years. Both are scheduled to hit the market in 2022.