Fei has over 10 years of development experience at Tencent and Sogou. Through this account, he shares his deep insights on performance.
This article dissects NVIDIA's Pascal architecture via the Tesla P100, detailing its 16nm process, HBM2 memory delivering 732 GB/s bandwidth, FP16 support via FP32 units achieving 21.2 TFLOPS, and NVLink 1.0 providing 160 GB/s GPU interconnect bandwidth, with comparative calculations across GPU generations.
The article analyzes NVIDIA's Maxwell GM200 architecture, detailing how splitting each SM into four independent processing blocks improved scheduling efficiency and core utilization, boosting FP32 performance to 6.84 TFLOPS on the Tesla M40 — a 20x increase over 10 years — while highlighting limited FP64 capability and memory bandwidth scaling challenges.
This article introduces remote-claude, an open‑source tool that lets you monitor and command Claude Code via a Feishu bot on your mobile device, detailing installation, Feishu bot setup, and step‑by‑step usage to enable hands‑free AI coding.
The article reflects on the evolution from the Internet boom to the AI surge, presents data on AI adoption and market impact, and offers seven concrete strategies—accepting AI, adopting proactive thinking, leveraging AI for expertise, diversifying production factors, boosting cognition, considering public service, and pursuing passion—to help individuals navigate and succeed in the AI-driven future.
This article explains how Linux processes and glibc threads allocate stacks, then details Go's runtime mechanisms for creating thread stacks and lightweight goroutine stacks, including guard‑based detection and automatic stack growth, with code examples and diagrams.
This article explains how to locate and monitor kernel execution using IDE navigation, Linux tracing tools such as ftrace, perf, tracepoints, and kprobes, and dives into the underlying implementation details of static and dynamic tracing mechanisms.
This article explains how Kubernetes uses both requests and limits to manage container CPU resources, detailing the underlying Linux cgroup mechanisms, the concept of over‑selling CPU cores, and practical guidance for SREs configuring optimal request and limit values.
This article explains Linux’s Completely Fair Scheduler, its data structures, how weight‑based CPU allocation works for containers via cgroup v1 and v2, and demonstrates the calculation of vruntime and practical examples of resource distribution across services.
This article examines NVIDIA's Kepler GPU architecture, highlighting its 28 nm process, increased transistor count, expanded CUDA core count, PCIe 3.0 support, enhanced memory hierarchy, new compute units, scheduling improvements like Hyper‑Q, and performance metrics of the Tesla K20X, illustrating the substantial gains over previous generations.
This article explains why modern enterprises increasingly need specialized performance engineering teams, outlines their ROI through cost savings, latency reduction, scalability, and engineering efficiency, details the engineers' responsibilities, and provides practical hiring guidelines and real‑world case studies.
