Russian scientific-technical center “Module” announced the beginning of mass production of the first Russian neural processor Module NM6408. For its development were spent 480 million rubles, and the market value of new items below foreign analogues. In addition, the chip surpassed the competitors in working with neural networks.

The production Module NM6408

Production of the new processor Module NM6408 is in Taiwan, but the company plans to transfer it to Russia after the emergence of the country’s factories, working on 28 nm technological process. The company representatives claim that their chip has no direct competition on the domestic market, as they are currently the only country in this region.

Cost and competitors

The cost Module NM6408 is only $70, while foreign competitors with similar characteristics is approximately $100. In the company of one of the competitors called the NVIDIA Jetson Xavier with similar characteristics and performance. At the same time, the Russian chip is the best performance of neural networks at a lower hardware cost with reusable superiority efficiency.

Structure and performance

Module NM6408 contains 21 core brand architecture of the NeuroMatrix Core with 28 nm technological process. The maximum level of performance is 512 GFlops. The chip was designed to handle large amounts of data in real time from a variety of fields. The new processor has already been developed by the computational module МС127.04 for embedded computing, specialized high-performance systems or to build platforms for digital signal processing and computer vision.

Modes of operation

A distinctive feature of the Module NM6408 is a function of switching between the two modes — 32 bits single-precision and performance of 512 GFlops, and the 64 bit double precision and 128 GFlops, respectively.

The head of Xiaomi revealed the price and features of Redmi 7
Apple acknowledged the problem with animation in iPhone Xs and Xs Max
Camera Redmi Note Pro 7 tested in real conditions

Read more •••

LEAVE A REPLY

Please enter your comment!
Please enter your name here