Run GLM 5.2 on 2 MacBooks with 128gb on RDMA with DS by antirez (twitter.com)

In a significant advancement for the DwarfStar community, developer antirez has successfully implemented DeepSeek v4 Flash and GLM 5.2 using Tensor Parallelism on two M5Max MacBooks, each equipped with 128GB of RAM, connected through Remote Direct Memory Access (RDMA). This breakthrough allows users to run GLM 5.2, a large language model, on consumer-grade hardware that previously struggled to accommodate its full structure due to memory constraints. This development is particularly noteworthy for the AI and machine learning community, as it lowers the barrier for accessing advanced language models, enabling more developers and researchers to experiment with and leverage these complex systems without investing in expensive infrastructure. The use of RDMA enhances performance by allowing faster data transfers between the two machines, making distributed model training more efficient. Consequently, this innovation not only democratizes access to powerful AI tools but also opens new avenues for collaboration and experimentation in the field of AI.