Faster memory reclamation with DPDK RCU
summary
Compares DPDK’s new RCU library with the Userspace RCU library by presenting benchmark tests that measure reclamation speed and behavior with pre-emptive readers. Describes DPDK RCU’s reclamation method, shows it avoids a dedicated background reclaim thread, integrates into existing libraries without app changes, and demonstrates significantly faster reclamation and better performance under pre-emption.
AI-written summary · about
related talks
28mCeph RBD Integration with DPUs
28mDPDK vs ODP: A comparison
28mPersistent Memory plus RDMA, new age remote device
14mODRP: On-Demand Remote Paging with Programmable RDMA
20mPeRF: Preemption-enabled RDMA Framework
35mIntroducing UACCE Bus of DPDK
20mMemory Recycling for Network Interfaces
54mPDCP Packet Processing Library in DPDK - Libpdcp