Nvidia’s next-generation Blackwell architecture GeForce RTX 50 GPUs will use TSMC’s most advanced node and support the DisplayPort 2.1 standard.According to the claim of Kopite7kimi, a user known for his closeness to Nvidia on Twitter (X), the new GPUs will be produced with TSMC’s 3nm production process.
There will be a significant increase in performance and efficiency
The new production process offers a 25-30% reduction in power consumption, a 10-15% increase in performance per transistor, a 42% smaller volume and a 1.7 times increase in density compared to the 4nm process used by Nvidia in its current GPUs. These significant gains could bring Nvidia significant gains in performance per watt over the GeForce RTX 40 “Ada Lovelace” series. At the same time, it can provide a significant advantage in competition with AMD.
It will come with GDDR7 memory
Nvidia’s HPC (High performance computing) and artificial intelligence-focused Blackwell GPUs will be produced with the 3nm process, like HBM3e memories. Meanwhile, it is claimed that Nvidia’s gaming-focused cards will come with GDDR7 memory. The Blackwell family is developed with a focus on both gaming and HPC/AI.
On the other hand, just like AMD’s Radeon RX 7000 series, Nvidia plans to offer DisplayPort 2.1 technology on its cards. GeForce RTX 50 Blackwell graphics processors are expected to be introduced between the end of 2024 and the beginning of 2025.
22768
