Use of PCIE and Clustering in AI and LLM Environments.

Abstract

Peripheral component Interconnect Express (PCIE) is an interconnect standard implemented as a tree network for connecting to peripheral devices to central processing units (CPU)s. Currently PCIE has become a basis for high speed interconnect applications. U.S. Pat. No. 11,194,754 describes a PCIE based interconnect scheme for data transport between the connected clusters of processing systems. PCIE interconnect using PCIE protocol is proposed for data transfer within and between the clusters of interconnected processors, memory modules, memory controllers, storage and other network components used in artificial intelligence (AI) systems with machine learning (ML) in large language model environments. The proposed architecture eliminates protocol changes and associated latency allowing high speed, high volume data transfer. The capability for remote direct memory access (RDMA) and direct memory access (DMA) capability provided by such PCIE based clusters without CPU involvement further enhances the speed of Data transfer.

Dependencies are inferred from the claim text.