Developer Weekly will select Top100 technical community/team blog’s weekly high-quality content and recommend it to CSDN developers. The content is mainly high-quality technical articles/videos and selected activities sharing, and it will be pushed every Friday. Welcome to follow and subscribe.
[Top quality content recommendation this week]
Dry | C / C ++ class should use the theoretical basis for migration
Lecturer | Han Guang, Senior Engineer of Loongson
Source| Loongson Technology Community
Abstract: The sharing of the first phase of the Loongson Ecological Forum will mainly explain from three aspects: the first aspect: the nature of application migration: why migration can be done on the Loongson computer, and the theoretical basis of application migration; the second aspect: to be migrated Application software classification: the classification of application software is explained from different aspects, which types of applications we use on Loongson Computer; the third aspect: application software migration plan: the basic steps and some technical details of application migration
Tags: C/C++, application migration
Release time: 2021-03-22
Source | Elastic China Community
Summary: Elastic version 7.12 is now officially launched on Elastic Cloud. This is the only managed Elasticsearch product that contains all the new features of the latest version. Now you can download Elastic Stack and cloud orchestration products (Elastic Cloud Enterprise and Elastic Cloud for Kubernetes). Self-managed deployment. This article will take you to explore the key highlights of this version.
Labels: Elastic, Big Data
Release time: 2021-03-25
Scenario-based active experiment of ByteDance chaos engineering practice
Source | ByteDance Technical Team Blog
Abstract: Distributed and microservices have become mainstream system architecture design schemes, and the availability of large-scale distributed systems has become the focus of attention. Chaos engineering has also begun to sprout and grow in major companies, but most of them are still in the initial stage of exploration. Various problems encountered in the process of practice have created resistance to the rapid implementation of Chaos Engineering. This article will introduce a key stage of ByteDance in the practice of chaos engineering: scenario-based active experiments, hoping to help you deepen your understanding of the value of chaos engineering, and provide more information for designing chaos engineering experiments and landing chaos engineering construction Ideas.
Tags: big data, distributed, artificial intelligence
Release time: 2021-03-25
Toutiao ANR Optimization Practice Series-Monitoring Tools and Analysis Ideas
Author | ByteDance Android Platform Architecture Team
Source | ByteDance Technical Team Blog
Abstract: Relying on the existing system logs is not enough to complete the problem attribution of complex scenarios, and some information cannot be obtained from the application side, which makes many online problems more difficult; therefore, we explored new monitoring capabilities on the application side. In order to make up for the shortcomings of insufficient information acquisition; at the same time, the log information and analysis ideas used in the daily analysis process are summarized to help you better master the analysis skills. This article can take you to take a look at the relevant implementation.
Tags: Android, ANR
Release time: 2021-03-23
ICLR 2021 | Meituan AutoML Paper: Robust Neural Network Architecture Search DARTS-
Author | Xiangxiang, Hyosung, Zhang Bo, Xiaolin, etc., all from the Meituan Visual Intelligence Center
Source| Meituan Technical Team
Abstract: The design and update iteration of high-quality models are the pain points and difficulties of current AI production and development. Under this background, automated machine learning (AutoML) has emerged. In 2017, Google officially proposed Neural Architecture Search (NAS) and became a core component of AutoML. The Meituan technical team has also actively explored the field of AutoML. This article is an interpretation of a paper that was included in the ICLR 2021 of the in-depth learning conference ICLR 2021 by the cooperation between Meituan and Shanghai Jiaotong University.
Tags: artificial intelligence, automated machine learning
Release time: 2021-03-25
Tracking the bald Crash, ten times the programmer’s Debug routine
Author | Tao Jianhui, founder of Taosi Data
Source| Taos Data Community
Abstract: When writing C programs, Crash is often encountered. In most cases, it is caused by null pointers or stray pointers. From the call stack, it is generally easy to find out the problem. But there is a type of Crash that is difficult to debug, that is, memory overflow. The overflowed part of the memory space just overwrites the data accessed by another thread (such as a structure), then when another thread reads this piece of data, the data obtained is invalid, which often leads to unforeseen errors or even crashes. Because the thread that caused the data overflow has left the scene, the problem is difficult to locate. This is an internal blog I wrote. Taking an error I encountered at the time as an example, I will share the methods to solve this kind of problem for your reference.
Labels: C, Crash
Release time: 2021-03-19
How to send notifications for machine learning exceptions
Source | Elastic China Community Official Blog
Abstract: For data, light collection is not enough. Faced with a large amount of data, we hope to discover events from these large amounts of data, which are different from usual events. We can use the machine learning provided by Elastic Stack to analyze data and generate abnormal events. Even so, abnormal events are obtained through the method of machine learning, but if they cannot be fed back to the monitoring person in time, these abnormal events cannot be processed in time, which will also affect the operation of the system. If these exceptions cannot be handled in time, it is fatal to the operation and maintenance of some systems. When machine learning helps us analyze anomalies, we can use alerts or watchers to notify them. This article will describe in detail how to deal with the exceptions that occur in machine learning.
Labels: machine learning, exception handling, data
Release time: 2021-03-25
Explain the profound things in a simple way | Talk about MNN GPU performance optimization strategy
Author | Tian Bu
Source| Alibaba Tao Department Technical Team Official Website Blog
Abstract: MNN (Mobile Neural Network) is a high-performance, general-purpose deep learning framework that supports efficient operation on various devices such as mobile, PC, server, and embedded devices. It uses the GPU capabilities of the device to fully "squeeze" The GPU resources of the “dry” device are used for high-performance deployment and training of deep learning. Since open source, MNN has been known for its high performance, versatility, ease of use and other features. In the past year, MNN GPU has renewed its efforts, and the OpenCL backend has improved overall performance for mobile (Adreno/Mali GPU) and PC by more than 100. %, the performance of some models doubled several times.
Tags: deep learning, artificial intelligence
Release time: 2021-03-24
Detailed low-latency high-quality sound: sound US- Hua and spatial sound articles
Source | Agora Technology Blog
Abstract: With the popularization of 5G and the update of codec technology, high-fidelity transmission of sound is no longer a problem. However, as more and more communication is moved online, people's demand for audio has changed from traditional listening to higher demands such as good sound and scene space restoration. This time we will talk about how we can make the sound better and more spacious in the real-time interactive scene. Finally, there are some small demos related to audio scenes, you can try them on your own.
Tags: artificial intelligence, sound beautification
Release time: 2021-03-23
Review the entire release of Zhiyuan Enlightenment 1.0
Source | Zhiyuan Community
Abstract: On March 20, Beijing Zhiyuan Artificial Intelligence Research Institute released "Enlightenment 1.0", which is my country's first ultra-large-scale intelligent model system. "Enlightenment 1.0" is led by the academic deputy dean of Zhiyuan Research Institute and Professor Tang Jie of Tsinghua University, leading a team of more than 100 AI scientists to jointly tackle key problems, and has achieved a number of internationally leading AI technology breakthroughs, forming a super-large-scale intelligent model training technology system , Trained a series of models including Chinese, multi-modality, cognition, and protein prediction.
Tags: artificial intelligence, sound beautification
Release time: 2021-03-25
Digital technology strategy: developer experience - the "last mile" of internal tools
Source| Developer Ecological Hero Club
Abstract: Domestic IT companies lack systematic thinking about "Developer Experience". The developer experience is similar to the user experience, except that the target is software developers. Corresponding to the international world is the perception and response of developers to the products, systems, or services that they use or expect to use. The difference is that the content that users pay attention to becomes the developer experience of libraries, SDKs, documents, frameworks, open source solutions, common tools, APIs, etc.
Label: Developer Experience
Release time: 2021-03-22
If you don’t understand, ask: How does the ZooKeeper cluster perform data synchronization?
Author | HelloGitHub-Old Xun
Source | Shave Weihan
Abstract: This article will take you to learn how Follower or Observer synchronizes data with Leader after the election, and what exactly is the authority management ACL provided by ZK.
Tags: database, ZooKeeper
Release time: 2021-03-25
[Recommended activities this week]
[Millions of people support open source] Looking for excellent open source projects , millions of programmers will help you!
Cooperative Community/Technical Blog
Alibaba Tao Department Technical Team Official Website Blog
Acoustic Network Agora Technology Blog