https://www.chenshaowen.com/en/tags/optimization/Recent content in Optimization on Shaowen Chen's WebsiteHugo -- gohugo.ioen©2016 - {year}, All Rights Reserved.Sat, 06 Sep 2025 00:00:00 +0000weeklyhttps://www.chenshaowen.com/en/blog/what-is-model-quantization.htmlSat, 06 Sep 2025 00:00:00 +0000Sat, 06 Sep 2025 00:00:00 +0000https://www.chenshaowen.com/en/blog/what-is-model-quantization.html1. What Is Model Quantization Model quantization is the process of converting the weights and activations of a high-precision model (usually 32-bit floating point FP32 or 16-bit floating point FP16) into a low-precision model (such as 8-bit integer INT8). The value range of FP32 is -3.4*10^38 to 3.4*10^38, with 4 billion values.微信公众号AILLMModel QuantizationOptimizationOperationshttps://www.chenshaowen.com/en/blog/custom-cluster-scheduler-to-optimize-tekton.htmlWed, 26 Apr 2023 10:00:00 +0000Wed, 26 Apr 2023 10:00:00 +0000https://www.chenshaowen.com/en/blog/custom-cluster-scheduler-to-optimize-tekton.html1. A Constrained Build Environment Cannot Meet Build Requirements Tekton is a CICD engine based on a Kubernetes cluster, and it is more cloud-native than Jenkins. In plain terms, that means it is easier to develop plugins for, easier to scale, easier to observe, and more fun. Because code can only be stored on the company intranet, the build cluster can only be deployed on the office intranet.微信公众号OptimizationTektonCICDDevOpsKubernetesOperationsLearningSchedulingBuildMulti-Clusterhttps://www.chenshaowen.com/en/blog/tekton-stress-test-and-optimize-k8s-cluster.htmlThu, 10 Nov 2022 00:00:00 +0000Thu, 10 Nov 2022 00:00:00 +0000https://www.chenshaowen.com/en/blog/tekton-stress-test-and-optimize-k8s-cluster.html1. Purpose of the Test Tune the parameters of the build cluster Probe the upper limit on the number of concurrent Tekton pipelines Give the optimal concurrency limit for a single cluster 2. Related Components and Machine Configuration Kubernetes version v1.21.4 Tekton version v0.24.1, kept consistent with production OpenEBS version localpv version 3.微信公众号TektonKubernetesStress TestingLarge ClustersOptimizationCICDOperationshttps://www.chenshaowen.com/en/blog/how-to-cache-third-party-package-whilie-building-in-stage.htmlTue, 22 Feb 2022 00:00:00 +0000Tue, 22 Feb 2022 00:00:00 +0000https://www.chenshaowen.com/en/blog/how-to-cache-third-party-package-whilie-building-in-stage.htmlIn a non-staged build scenario, when we use a container to build, we can mount the cache directory inside the container onto the build host, run the build task, and then copy the artifacts into the runtime image to produce the application image. But with a staged build, the build image and the runtime image live in the same Dockerfile, which makes it harder to optimize the caching of third-party dependencies.微信公众号CICDCIDevOpsCacheOptimizationhttps://www.chenshaowen.com/en/blog/practical-algorithm-of-bloom-filter.htmlTue, 01 Dec 2020 00:00:00 +0000Tue, 01 Dec 2020 00:00:00 +0000https://www.chenshaowen.com/en/blog/practical-algorithm-of-bloom-filter.html1. What Is Bloom Filter A Bloom Filter is a data structure proposed by Bloom in 1970. It maps elements (x, y, z) through a series of functions into a binary vector (a 0101 sequence), and is used to quickly determine whether an element w is in a set. As shown below (from Wikipedia):微信公众号AlgorithmsR&DGoOptimizationBloom Filterhttps://www.chenshaowen.com/en/blog/database-query-optimization-of-django-performance.htmlWed, 20 Mar 2019 00:00:00 +0000Wed, 20 Mar 2019 00:00:00 +0000https://www.chenshaowen.com/en/blog/database-query-optimization-of-django-performance.htmlThis article mainly offers optimization advice on Django fields and queries, and also introduces a performance analysis tool called Django-silk. I hope it helps you develop high-performance Django projects. 1. DBA’s Advice 1.1 Table Field Design Avoid null values; null values are hard to optimize queries around and take up extra index space Prefer INT over BIGINT, and describe fields as accurately as possible Use enums or integers instead of string types Use TIMESTAMP instead of DATETIME Do not put more than 20 fields in a single table Store IPs as integers 1.微信公众号DatabaseOptimizationPerformanceDjangoPythonR&Dhttps://www.chenshaowen.com/en/blog/django-full-stack-optimization-guide.htmlThu, 15 Nov 2018 00:00:00 +0000Thu, 15 Nov 2018 00:00:00 +0000https://www.chenshaowen.com/en/blog/django-full-stack-optimization-guide.htmlAs the data volume exploded, the system responded very slowly. We carried out a series of optimizations on the application, and the system’s response time improved by an order of magnitude. Overall, optimizations in file compression and faster network access gave a noticeable boost to frontend performance, while optimizations in stored procedures, caching, and logic code gave a noticeable boost to backend performance.微信公众号DjangoPythonOptimizationFrontendBackendGuideR&Dhttps://www.chenshaowen.com/en/blog/ci-script-of-how-to-use-cdn-to-separate-frontend-and-backend-more.htmlSat, 15 Sep 2018 00:00:00 +0000Sat, 15 Sep 2018 00:00:00 +0000https://www.chenshaowen.com/en/blog/ci-script-of-how-to-use-cdn-to-separate-frontend-and-backend-more.htmlIn a team, any change to the development workflow must be paired with the corresponding automation tooling. Without a low enough cost of use, the change is meaningless, because nobody will actually use it. In the previous post we covered how to use a CDN for further frontend/backend separation; this微信公众号CICDNOptimizationDeploymentPaaSScriptingDevOpsR&D