With a powerful 32 petaFLOPS of AI compute for the IPU-POD 128 and 64 petaFLOPS for the IPU-POD 256, Graphcore said that it would now be able to extend its reach into AI supercomputer territory.
Both systems have been developed for cloud hyperscalers, national scientific computing labs and enterprise companies with large AI teams in markets like financial services or pharmaceutical.
The IPU-PODs enable, for example, faster training of large Transformer-based language models across an entire system, running large-scale commercial AI inference applications in production, giving more developers IPU access by dividing up the system into smaller, flexible vPODs or enabling scientific breakthroughs by enabling exploration of new and emerging models like GPT and GNNs across complete systems.
Graphcore is able to provide extensive training and support to help customers accelerate time to value from IPU-based AI deployments.
As with the company's other IPU-POD systems, the disaggregation of AI compute and servers means that both the 128 and the 256 can be optimised to deliver maximum performance for different AI workloads. For example, an NLP-focused system could use as few as two servers for IPU-POD 128, while a more data-intensive task such as computer vision tasks may benefit from an eight-server setup.
Additionally, system storage can be optimised around particular AI workloads, using technology from Graphcore’s storage partners.
