The “PEZY-SC” series, our self-developed domestic processor, now supports the AI framework “PyTorch”. This will operate leading-edge AI applications more efficiently on the fourth-generation PEZY-SC4S large-scale microprocessor, which is scheduled for release at the end of the year.

Currently, the following major AI frameworks, libraries, and models were confirmed to work on our ZettaScaler-3.0 supercomputer, which is equipped with four PEZY-SC3 units.

  Frameworks & Libraries: PyTorch, Accelerate, DeepSpeed, Transformers, vLLM, Text Generation Inference, Diffusers

  Models: Gemma3, Llama3, Qwen2, StableDiffusion2, HuBERT, VisionTransformer

We are strongly promoting the utilization of generative AI in the healthcare field by leveraging our long experience in developing “medical diagnostic imaging devices” and “genome analysis systems”. We will run various AI applications on the next-generation supercomputer “ZettaScaler-4.0” equipped with “PEZY-SC4S”, which is a domestic GPGPU, to improve the accuracy of diagnosis and operational efficiency in the medical field, and to accelerate research and development.