Condor Galaxy 1, a 54 Million Core Cloud AI Supercomputer
Condor Galaxy 1 (CG-1) is an AI supercomputer that delivers 4 exaFLOPS of AI compute at FP16. One of the largest AI supercomputers in production, CG-1 is comprised of 64 Cerebras CS-2 systems, linked together via Cerebras SwarmX technology. Supporting up to 600 billion parameters, and extensible up to 100 trillion, CG-1 exhibits near-perfect linear scaling and runs strictly data parallel, so you spend no time on distributed compute.
Operated by Cerebras as part of the Cerebras Cloud, CG-1 is the first of three US-based AI supercomputers to be built in partnership with G42, the UAE-based AI pioneer. After completing CG-1, CG-2 and CG-3, we plan to build an additional six, 4 exaFLOP AI Supercomputers with G42, for a total of 36 exaFLOPS of interconnected AI compute. This constellation of nine AI supercomputers is expected to be completed in 2024.
CG-1 is available through the Cerebras Cloud via our White Glove service for your most challenging AI workloads.
By the numbers
CG-1 is the largest Cerebras Wafer-Scale Cluster in operation today. The raw statistics are impressive; the performance is incredible.
82
terabytes of memory
600
billion parameter models, extensible to 100 trillion
388
terabits-per-second node-node fabric bandwidth
72,704
AMD Epyc™ Gen 3 cores
Unprecedented ease-of-use
Near-Perfect Performance Scaling
Unlike any known GPU-based cluster, Andromeda delivers near-perfect scaling across GPT-class large language models, including GPT variants, Falcon and Llama. Near-perfect scaling means that that as additional CS-2s are used, training time is reduced in near perfect proportion.
Setting up a generative AI model takes minutes, not months and can be done by a single person.
What our customers are saying
Get a Test Drive
If you would like to experience the power and ease of AI Supercomputers from Cerebras, access to our 1 exaFLOP AI supercomputer, Andromeda, and the 4 exaflop CG-1, is available now through Cerebras Cloud.