Brain systems known for building massive computing clusters used for all kinds of AI and scientific tasks. has once again broken records in the AI industry by unveiling its latest technological marvel, the Unit Scale Engine 3 (WSE-3)), billed as the fastest AI chip the world has seen to date. With an astonishing 4 trillion transistors, this chip is designed to power the next generation of AI supercomputers, delivering unprecedented levels of performance.
The WSE-3, designed using a state-of-the-art 5nm process, forms the backbone of the Cerebras CS-3 AI supercomputer. It delivers a revolutionary peak AI performance of 125 petaflops, enabled by 900,000 AI-optimized compute cores. This development marks a significant step forward, doubling the performance of its predecessor, the WSE-2, without increasing power consumption or cost.
Cerebras Systems’ ambition to revolutionize AI computing is evident in the WSE-3 Specifications. The chip has 44GB of onboard SRAM and supports external memory configurations ranging from 1.5TB to a colossal 1.2PB. This vast memory capacity enables the training of AI models ranging in size up to 24 trillion parameters, facilitating the development of models ten times larger than those like GPT-4 and Gemini.
One of the most compelling aspects of the CS-3 is its scalability. The system can be clustered up to 2,048 CS-3 units, achieving a staggering 256 exaFLOPs of computing power. This scalability isn’t just about raw power; it simplifies the AI training workflow, improving developer productivity by enabling the training of large models without the need for complex partitioning or refactoring.
Cerebras’ commitment to advancing AI technology extends to its software framework, which now supports PyTorch 2.0 and the latest AI models and techniques. This includes native hardware acceleration for dynamic, unstructured sparsity, which can speed up training times up to eight times.
Cerebras’ journey, as told by CEO Andrew Feldman, from skepticism eight years ago to the launch of the WSE-3, embodies the company’s pioneering spirit and commitment to pushing the boundaries of AI.
“When we started this journey eight years ago, everyone said that wafer-scale processors were a pipe dream. We couldn’t be more proud to introduce the third generation of our revolutionary wafer-scale AI chip,» said Andrew Feldman, CEO and co-founder of Cerebras. “WSE-3 is the world’s fastest AI chip, purpose-built for the latest cutting-edge AI work, ranging from expert blending to 24 trillion parameter models. We are excited to bring WSE-3 and CS-3 to market to help solve today’s biggest challenges in AI.»
This innovation has not gone unnoticed, with a significant backlog of orders for the CS-3 from international businesses, government entities and clouds. The impact of Cerebras’ technology is further highlighted through strategic partnerships, such as with G42which led to the creation of some of the largest AI supercomputers in the world.
As Cerebras Systems continues to pave the way for future advancements in AI, the launch of the WSE-3 is a testament to the incredible potential of wafer-scale engineering. This chip is not just a piece of technology; it is a gateway to a future where the boundaries of AI are continually pushed, promising new possibilities for research, enterprise applications and beyond.
