Processing-In-Memory Primitives for Data Management (PIMPMe)

• Applicants: Dr.-Ing. Andreas Becher; Professor Dr.-Ing. Kai-Uwe Sattler; Professor Dr.-Ing. Daniel Ziener
• Subject Area: Data Management, Data-Intensive Systems, Computer Science Methods in Business Informatics
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 502352642

Ongoing developments in the field of memory technology open up great potential for building highly efficient data-intensive systems addressing the challenges of modern applications. At the same time, it requires to rethink traditional assumptions to fully exploit this technology. In this project we plan to investigate the benefit of memory-centric computing paradigms: we will conduct research on offloading computation to memory in order to reduce the amount of data to be transferred between memory and CPU and in this way increase bandwidth, to reduce the CPU load for computation, and in the end, help to reduce the energy consumption of modern IT systems. We plan to particularly extend the results of the first project phase on offloading database computation to memory. Specifically, we plan the following two dimensions of extension: (1) PIM abstractions: While in the first phase we considered primitives tailored for the UPMEM memory technology and are still working on initial FPGA support, in the second phase we plan to design and develop higher-level abstractions that support different storage and execution models. (2) Distributed computing: With the advent of interconnects such as CXL.mem, but also feasible with networking technologies such as RDMA, shared memory infrastructures are becoming available where multiple compute nodes can access the same memory and offload computation. We plan to explore the challenges and opportunities of such infrastructures in the specific context of distributed database processing.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2377: Disruptive Main-Memory Technologies