Huawei Puts Pressure on Nvidia with Efficient Chip-Scale Architecture

Huawei's 18-tier pagoda architecture is challenging Nvidia's chip-scaling playbook, according to industry sources. This approach allows for more efficient use of silicon and could be a game-changer in the field of high-performance computing.

The Huawei pagoda architecture uses a novel technique called 'chip scaling', which enables the company to pack more transistors into a smaller space than traditional methods allow. This means that Huawei's chips can offer higher performance at lower power consumption, making them more attractive to customers.

Nvidia's current approach to chip scaling is based on using a combination of 3D stacked transistors and FinFET technology. However, the company's use of these techniques has been criticized for being inefficient and leading to excessive power consumption.