0 views collected around this technical thread.
This article explains common interview questions on frequent Full GC and high CPU usage in Java backend development, detailing root causes, JVM parameter adjustments, memory leak prevention, and code optimization techniques, along with practical code examples and tool recommendations for effective performance tuning.
The article details MNN’s CPU backend dynamic quantization for Transformer‑type models, describing runtime int8 conversion, block‑wise matrix‑multiply optimizations using ARM SMMLA/SDOT and AVX‑512 VNNI, weight‑group and batch‑wise quantization techniques, and reports up to three‑fold speed‑ups on Snapdragon 8 Gen 3.
Learn how to identify and resolve extreme CPU usage spikes—up to 900%—in MySQL and Java applications by using top, show processlist, indexing, caching, thread analysis, and code adjustments, with real‑world examples and step‑by‑step command guides.
This article explains how to evaluate device overheating using metrics such as CPU/GPU usage and temperature, collect and visualize Trace data with SmartPerf or DevEco Studio, analyze high‑load threads and frequency scaling, and apply targeted code changes—disabling sensors, releasing Lottie resources, and simplifying components—to reduce heat and restore frame‑rate stability.
This article walks through a real‑world incident where a data‑platform server showed near‑100% CPU usage, explains step‑by‑step investigation using top, pwdx, and jstack, identifies a time‑conversion utility as the root cause, and presents a streamlined script‑based solution that reduced CPU load by thirtyfold.
PP-LCNet is a lightweight convolutional neural network designed for Intel CPUs that leverages MKLDNN acceleration, H‑Swish activation, selective SE modules, larger kernels, and expanded fully‑connected layers to achieve higher accuracy without increasing inference latency across image classification, detection, and segmentation tasks.
This article explains how to diagnose and resolve severe CPU usage spikes—up to 900%—in MySQL and Java processes, detailing step‑by‑step identification, common causes such as missing indexes or excessive garbage collection, and practical remediation techniques including query optimization, indexing, caching, and thread analysis.
This guide explains how to identify and fix extreme CPU usage—up to 900%—in MySQL and Java processes, covering diagnostic commands, SQL indexing, caching strategies, thread analysis, and code adjustments to stabilize production systems.
This guide explains how high concurrency and fast response depend on throughput and latency, defines key performance metrics, shows how to interpret average load, CPU context switches, and memory usage, and provides practical Linux tools and command‑line examples for diagnosing and tuning system performance.
The article explains how to boost Android app performance by programmatically raising CPU frequency using Qualcomm’s BoostFramework, setting thread‑core affinity with native sched_setaffinity calls, and increasing execution priority via Process.setThreadPriority, providing Kotlin and C examples that demonstrate measurable speed gains.
This article explains the design of a compact RPC service that aggregates product benefit information, detailing how Sirector thread scheduling reduces CPU consumption and how abstract handlers and Spring prototype beans improve code maintainability and extensibility.
The article explains that CPU spikes during Java deployments are caused mainly by HotSpot C2 JIT compiler threads, and demonstrates that tiered compilation, CodeCache sizing, JWarmUp pre‑warming, gradual traffic ramp‑up, and higher JIT compile thresholds together effectively smooth startup CPU usage.
This article explains how hardware architecture, Linux OS version, kernel configuration, OVS integration, memory management, NUMA awareness, and CPU micro‑architecture affect DPDK application performance and provides concrete tuning steps such as CPU isolation, service disabling, huge‑page setup, and optimized memory allocation.
This article demonstrates how adding two gevent monkey‑patch lines to a feapder spider reduces CPU usage from 121% to 99% while changing the effective thread count from 36 to 12, and discusses the underlying principle, performance trade‑offs, and future directions for coroutine support.
The paper presents a unified, high‑performance frame‑extraction framework that dynamically selects CPU or GPU decoding, leverages multithreaded and CUDA‑accelerated pipelines, keeps frames in memory, and achieves up to ten‑fold latency reductions for diverse AI video‑inference tasks.
The article demonstrates how Go’s built‑in pprof tools can be used for live performance profiling, walking through two real‑world cases—one where a malformed JSON request caused massive object allocation and CPU spikes, and another where per‑call self‑referencing structs leaked memory—while offering practical tips on input validation, allocation reduction, and GC monitoring.
This article explains which components (user processes, IO wait, interrupts) consume most CPU in MySQL workloads, the performance impacts of high CPU usage, and practical steps—reducing IO, optimizing queries, and hardware upgrades—to lower CPU consumption and improve throughput.
The article analyzes the growing demands on network I/O, explains Linux and x86 bottlenecks, introduces DPDK’s user‑space bypass architecture and its core optimizations such as hugepages, poll‑mode drivers, SIMD, and CPU‑specific tuning, and finally discusses the DPDK ecosystem and practical considerations for backend developers.
To resolve a MySQL single‑core CPU bottleneck, examine and tune slow queries and indexes, eliminate heavy lock contention, break large transactions into smaller ones, and employ tools like perf top to pinpoint performance hotspots, as recommended by Tencent Cloud Database expert Liu Di.
This article explains how tuning the PHP‑FPM max_requests setting and modifying its source code can eliminate periodic CPU_IDLE and MEM_USED fluctuations in high‑traffic web services, resulting in smoother resource utilization and more reliable performance.