Cerebras systems presents the largest chip ever built

Cerebras Systems Builds is the largest processor ever built with 1.2 trillion transistors

The Cerebras Wafer Scale Engine has 1.2 trillion transistors, which are basic electronic on-off switches that are the building blocks of any silicon chip. Intel's first 4004 processor in 1971 had 2, 300 transistors, and a current AMD processor has 32 billion transistors, just to put this in perspective.

In this way, the Cerebras Wafer Scale Engine is the largest processor ever built, and has been specifically designed to process artificial intelligence applications. The company talks about design this week at the Hot Chips conference at Stanford University in Palo Alto, California.

The Cerebras chip has 400, 000 nuclei in 42, 225 square millimeters. It is 56.7 times larger than Nvidia's largest graphics processing unit, measuring 815 square millimeters and having 21.1 billion transistors.

The WSE also contains 3, 000 times more high-speed on-chip memory and has 10, 000 times more memory bandwidth. The amount of memory is 18 gigabytes SRAM.

Visit our guide on the best processors on the market

Chip size is profoundly important in AI, as large chips process information faster, producing responses in less time. Reducing time to understand, or "training time, " allows researchers to test more ideas, use more data, and solve new problems. Google, Facebook, OpenAI, Tencent, Baidu, and many others argue that the fundamental limitation of AI today is that it takes too long to train models. The reduction in training time thus eliminates a significant bottleneck for the progress of the entire industry.

The record accomplishments of the Cerebra WSE would not have been possible without years of close collaboration with TSMC. The WSE is manufactured by TSMC in its advanced 16nm process technology.

In this way, it seems that a very important step has been taken in the process for the processing of current and future AI.

Venturebeat font