👋
Welcome
to
Cuterwrite 's
Blog
OpenMP (Open Multi-Processing) is a widely used multithreading parallel programming model that provides a rich set of instructions and APIs for parallel computing on shared memory systems. Originating in 1997, OpenMP was standardized by multiple leading hardware and software vendors with the aim of simplifying the design and implementation process of parallel programs to fully utilize the computing power of modern multi-core processors. This article will introduce the basics and programming techniques of OpenMP.
This article is reprinted from Zhihu Column: 3. Basic Elements of RDMA, author: Savir. When discussing the RDMA protocol, various abbreviations are unavoidable. Unlike the original IB protocol, this article provides a simple and straightforward introduction to the most important queue concepts in RDMA, making it easier for readers to understand.
This article is reprinted from Zhihu Column: 2. Comparison between communication based on traditional Ethernet and RDMA technology, author: Savir. Traditional Ethernet provides a Socket interface as an API to upper-layer applications, while the RDMA protocol stack provides a verbs interface to the upper layer. There are significant differences between these two communication methods in the process of data reaching the memory space of the remote user from the local user's memory space. This article aims to explain to readers why we should use RDMA technology by comparing these differences.
Unified Communication X (UCX) and Open Message Passing Interface (OpenMPI) are two indispensable components in the field of high-performance computing. UCX provides an efficient set of low-level communication libraries, optimizing hardware resource utilization, while OpenMPI is a message passing interface widely used in parallel computing tasks. This article will provide detailed guidance on how to compile and install UCX version 1.15.0 from source code and the compatible OpenMPI version 5.0.0.
GCC-13 is a major update. In this version, GCC's static analyzer continues to improve, with more C23 and C++23 features, and supports many new x86_64/RISC-V/AArch64 processors. To promote the use of the new C++ standards instead of always writing C++17 or even C++11, I decided to compile and install GCC-13.2.0 to experience the new C++ standards.