Cerebras Systems Unveils World’s Fastest AI Chip with Whopping 4 Trillion Transistors

0
38
Cerebras Systems Unveils World’s Fastest AI Chip with Whopping 4 Trillion Transistors
Cerebras Systems Unveils World’s Fastest AI Chip with Whopping 4 Trillion Transistors

The Third Generation 5nm Wafer Scale Engine (WSE-3) Powers the Industry’s Most Scalable AI Supercomputers, Up To 256 exaFLOPs via 2048 Nodes

The latest Cerebras Software Framework provides native support for PyTorch 2.0 and the latest AI models and techniques, such as multi-modal models, vision transformers, mixtures of experts, and diffusion.

Bangalore, India, March 15, 2024: Cerebras Systems, the pioneer in accelerating generative AI, has doubled down on its existing world record for the fastest AI chip with the introduction of the Wafer Scale Engine 3. The WSE-3 delivers twice the performance of the previous record-holder, the Cerebras WSE-2, at the same power draw and for the same price. Purpose built for training the industry’s largest AI models, the 5nm-based, 4 trillion transistor WSE-3 powers the Cerebras CS-3 AI supercomputer, delivering 125 petaflops of peak AI performance through 900,000 AI-optimized compute cores.

Key Specs:

  • 4 trillion transistors
  • 900,000 AI cores
  • 125 petaflops of peak AI performance
  • 44GB of on-chip SRAM
  • 5nm TSMC process
  • External memory: 1.5TB, 12TB, or 1.2 PB
  • Trains AI models with up to 24 trillion parameters.
  • Cluster size of up to 2048 CS-3 systems

With a huge memory system of up to 1.2 petabytes, the CS-3 is designed to train next-generation frontier models 10x larger than the GPT-4 and Gemini. 24 trillion parameter models can be stored in a single logical memory space without partitioning or refactoring, dramatically simplifying the training workflow and accelerating developer productivity. Training a one-trillion-parameter model on the CS-3 is as straightforward as training a one-billion-parameter model on GPUs.

The CS-3 is built for both enterprise and hyperscale needs. Compact four-system configurations can fine-tune 70B models in a day, while at full scale using 2048 systems, Llama 70B can be trained from scratch in a single day—an unprecedented feat for generative AI.

The latest Cerebras Software Framework provides native support for PyTorch 2.0 and the latest AI models and techniques, such as multi-modal models, vision transformers, mixtures of experts, and diffusion. Cerebras remains the only platform that provides native hardware acceleration for dynamic and unstructured sparsity, speeding up training by up to 8x.

“When we started on this journey eight years ago, everyone said wafer-scale processors were a pipe dream. We could not be more proud to be introducing the third generation of our groundbreaking water-scale AI chip,” said Andrew Feldman, CEO and co-founder of Cerebras.

“WSE-3 is the fastest AI chip in the world, purpose-built for the latest cutting-edge AI work, from a mixture of experts to 24 trillion parameter models. We are thrilled to bring WSE-3 and CS-3 to market to help solve today’s biggest AI challenges.”

Superior power efficiency and software simplicity

With every component optimized for AI work, CS-3 delivers more compute performance in less space and less power than any other system. While GPU power consumption is doubling from generation to generation, the CS-3 doubles performance but stays within the same power envelope. The CS-3 offers superior ease of use, requiring 97% less code than GPUs for LLMs and the ability to train models ranging from 1B to 24T parameters in purely data-parallel mode. A standard implementation of a GPT-3-sized model required just 565 lines of code on Cerebras—an industry record.

Industry Partnerships and Customer Momentum

Cerebras already has a sizeable backlog of orders for CS-3 across enterprise, government, and international clouds.

“As a long-time partner of Cerebras, we are excited to see what’s possible with the evolution of wafer-scale engineering. CS-3 and the supercomputers based on this architecture are powering novel-scale systems that allow us to explore the limits of frontier AI and science,” said Rick Stevens, Argonne National Laboratory Associate Laboratory Director for Computing, Environment, and Life Sciences. “The audacity of what Cerebras is doing matches our ambition, and it matches how we think about the future.”

“As part of our multi-year strategic collaboration with Cerebras to develop AI models that improve patient outcomes and diagnoses, we are excited to see advancements being made on the technology capabilities to enhance our efforts,” said Dr. Matthew Callstrom, M.D., Mayo Clinic’s medical director for strategy and chair of radiology.

The CS-3 will also play an important role in the pioneering strategic partnership between Cerebras and G42. The Cerebras and G42 partnership has already delivered 8 exaFLOPs of AI supercomputer performance via Condor Galaxy 1 (CG-1) and Condor Galaxy 2 (CG-2). Both CG-1 and CG-2, deployed in California, are among the largest AI supercomputers in the world.

Today, Cerebras and G42 announced that Condor Galaxy 3 is under construction. Condor Galaxy 3 will be built with 64 CS-3 systems, producing 8 exaFLOPs of AI compute, making it one of the largest AI supercomputers in the world. Condor Galaxy 3 is the third installation in the Condor Galaxy network. The Cerebras G42 strategic partnership is set to deliver tens of exaFLOPs of AI compute. Condor Galaxy has trained some of the industry’s leading open-source models, including Jais-30B, Med42, Crystal-Coder-7B, and BTLM-3B-8K.

“Our strategic partnership with Cerebras has been instrumental in propelling innovation at G42 and will contribute to the acceleration of the AI revolution on a global scale,” said Kiril Evtimov, Group CTO of G42. “Condor Galaxy 3, our next AI supercomputer boasting 8 exaFLOPs, is currently under construction and will soon bring our system’s total production of AI compute to 16 exaFLOPs.”

For more information, please visit https://www.cerebras.net/product-system/.

Also readWomen in the technology industry is constantly increasing, says Rajita Bhatnagar

Do FollowCIO News LinkedIn Account | CIO News Facebook | CIO News Youtube | CIO News Twitter 

About us:

CIO News, a proprietary of Mercadeo, produces award-winning content and resources for IT leaders across any industry through print articles and recorded video interviews on topics in the technology sector such as Digital Transformation, Artificial Intelligence (AI), Machine Learning (ML), Cloud, Robotics, Cyber-security, Data, Analytics, SOC, SASE, among other technology topics.