NVIDIA: GPU architecture leads rivals for more than 2 years 12nm process will be upgraded

With the de facto cancellation of the GTC 2020 conference, the possibility of NVIDIA releasing the next-generation Ampere graphics card in March this year is largely out of sight. NVIDIA isn’t in a hurry on the issue, with senior vice president Jeff Fisher saying their GPU architecture is more than two years ahead of its rivals.

NVIDIA: GPU architecture leads rivals for more than 2 years 12nm process will be upgraded

In a recent conference call, Jeff Fisher, senior vice president of business at NVIDIA GeForce, spoke again about their situation in the game card market. In his view, the GPU architecture is more important than the process, and although NVIDIA’s current graphics card is still made in a 12nm process, they are more than two years ahead of their rival’s 7nm graphics card.

Considering that NVIDIA launched its 12nm Turing GPU in August 2018, it is still the only graphics card that supports hardware and optical pursuit technologies, and the performance and energy efficiency remain sway, and even with the launch of the big Navi flagship card this year, the performance will be only the RTX 2080 Ti class.

From the perspective of NVIDIA, it seems that there is really no need to rush to upgrade the process, the gap is now enough for the opponent to catch up. But Jeff Fisher did not put the word dead, after all, as the opponent again used the 7nm-plus process, the bilateral process gap is widening, NVIDIA also had to upgrade, he mentioned that the 12nm process will be abandoned sooner or later. Of course, when to give up, the next generation of ampere in the end use of what process will not be mentioned, after all, has not been released.

Combined with the previous revelations, NVIDIA’s ampere GPU upgrade 7nm is unlikely, should be the GA100 core is possible on the 7nm process, mainly used in data centers and other markets.

The core of the GA102 and below may be Samsung’s improved 8nm LPP process, after all, last year’s DRIVE AGX Orin chip was an 8nm LPP process contract.