Cerebras and G42 Complete 4 exaFLOP AI Supercomputer and Start the March Towards 8 exaFLOPs
With the completion of Condor Galaxy 1, Cerebras and G42 have broken ground on the Condor Galaxy 2 (CG-2).
Cerebras Systems, the pioneer in accelerating generative AI, and G42, the leading UAE-based technology holding group, today announced that the Condor Galaxy network of nine interconnected supercomputers has reached its second phase of build out. With the completion of Condor Galaxy 1, Cerebras and G42 have broken ground on the Condor Galaxy 2 (CG-2). CG-2 will be four exaFLOPs and 54 million AI-optimized compute cores, expanding the Condor Galaxy network to a total of eight exaFLOPs and 108 million cores upon completion. This represents an important milestone in Cerebras and G42’s plan to build a 36 exaFLOP constellation of AI supercomputers.
“G42’s strategic partnership with Cerebras Systems is paving the way toward our joint vision of creating the world’s largest and fastest AI supercomputers,” said Kiril Evtimov, Group CTO, G42. “We support organizations working with unique, massive, and diverse datasets that require extensive AI training capacity. These use cases span across different sectors including healthcare, energy, investments, and climate studies to name a few.”
“Since we started our collaboration, we successfully launched Jais 13B and 30B bilingual models and supported the release of Med42 Clinical LLM. These early successes illustrate the scale and ease of use of the Condor Galaxy platform. As a result, our rate of iteration from idea, to experiment, to solution has never been faster and has proven to be a game changer for us,” said Dr. Andrew Jackson, EVP, Chief AI Officer at Core42, a G42 company and the national-scale enabler for cloud and generative AI.
Training large AI models requires a massive amount of compute, vast datasets, and specialized AI expertise. The G42 and Cerebras partnership delivers on all three of these elements. By building and commercializing the largest AI supercomputers, the Condor Galaxy constellation enables simple and easy access to the largest and fastest AI compute in the world. The models trained on the Condor Galaxy have the potential to move humanity forward, making significant discoveries and breakthroughs.
“With CG-1 now complete, we're already seeing the impact and important contributions that this strategic partnership delivers,” says Andrew Feldman, CEO and co-founder, Cerebras Systems. “In partnership with G42, we are changing the worldwide inventory of compute and using our combined expertise to advance AI work in a powerful way, to quickly and efficiently train leading LLMs.”
Condor Galaxy is designed to enable G42 and its customers to train large, ground-breaking models quickly and easily, thereby accelerating the pace of innovation. The Cerebras-G42 strategic partnership has already advanced state-of-the-art AI models in healthcare with the introduction of Med42, the leading generative AI medical assistant, as well as brought Arabic Language chat to more than 400 million Arabic speakers through the introduction of Jais 30B, the premier Arabic language LLM. Pioneering work on CG-1 has also included AI-based climate studies as well as pioneering work in high-performance computing.
In addition to CG-1 and CG-2, Cerebras and G42 previously announced their plans to deploy seven additional supercomputers -- CG-3 through CG-9 -- bringing the total compute power of the Condor Galaxy to 36 exaFLOPs. This entails deploying 576 Cerebras CS-2 systems and feeding the cluster with more than 654,000 AMD CPU cores. With 36 exaFLOPs in total, this unprecedented constellation of supercomputers will revolutionize the advancement of AI globally. Access to CG-1 is available now.
About the Condor Galaxy 2
Optimized for Large Language Models (LLMs) and Generative AI, CG-2 will deliver 4 exaFLOPs of 16-bit AI compute, with standard support for up to 600 billion parameter models and extendable configurations that support up to 100 trillion parameter models. With 54 million AI-optimized compute cores, 388 terabits per second of fabric bandwidth, CG-2 is fed by 72,704 AMD EPYC processor cores. And, unlike any known GPU cluster, CG-2 delivers near-linear performance scaling from 1 to 64 CS-2 systems using simple data parallelism.
CG-2 is powered by the Cerebras Software Platform, which provides native support for PyTorch 2.0 and a wide family of models, such as GPT, Diffusion Transformer, T5, and U-Net. Support for new model variants are added in near real-time, with Llama 2, StarCoder, BLOOM, and Falcon being recent additions.
CG-2 is designed to train the largest models, with long context, out of the box. ML engineers can fine tune Llama 2 70B on a single machine or scale out to 64 nodes with no code changes. With 40GB of on-chip memory, CS-2 supports 50k context length natively. Training on CG-2 does not require complex model distribution frameworks such as Megatron or DeepSpeed, allowing developers to work on the largest models as if they had the simplicity of a 1B parameter model, streamlining development and shortening time to solution.
Each Cerebras CS-2 system is designed, packaged, manufactured, tested, and integrated in the U.S.; Cerebras is the only AI hardware company to package processors and manufacture AI systems in the U.S. CG-2 is located in California.
CG-2 is the second 4 exaFLOP AI supercomputer to be deployed in California by Cerebras and Core42.
About Cerebras Systems
Cerebras Systems is a team of pioneering computer architects, computer scientists, deep learning researchers, and engineers of all types. We have come together to build a new class of computer system, designed for the singular purpose of accelerating generative AI. Our flagship product, the CS-2 system, powered by the world’s largest and fastest AI processor, makes training large models simple and easy, by avoiding the complexity of distributed computing. Cerebras solutions are available in the cloud, through the Cerebras AI Model Studio or on premises. For further information, visit https://www.cerebras.net.
About G42
G42 is a global leader in creating visionary artificial intelligence capabilities for a better tomorrow. Born in Abu Dhabi and operating around the world, G42 champions AI as a powerful force for good. Its people are constantly reimagining what technology can do, applying advanced thinking and innovation to accelerate progress and tackle society’s most pressing problems.
G42 is joining forces with nations, corporations and individuals to create the infrastructure for tomorrow’s world. From molecular biology to space exploration and everything in between, G42 realizes exponential possibilities, today.
For further information visit www.g42.ai.
View source version on businesswire.com: https://www.businesswire.com/news/home/20231113436237/en/