1
AMD 发布 ROCm 10.0.0，重点关注 AI 推理、开发者工具、性能分析
IT之家 8 月 28 日消息，AMD 本周发布了 GPU 加速计算开放软件堆栈 ROCm 的 10.0.0 版本，这一版本号旨在纪念其诞生 10 周年。新版本 重点关注 Instinct、Radeon、Ryzen AI 平台上的 AI 推理、开发者工具、性能分析 。 ROCm 10.0.0 扩展了…
2
From Caveman to Expert Analyst: Energy Consumption of Variable LLM Tasks
从原始人式推理到专家级分析，看不同LLM任务能耗差异，为绿色AI提供量化视角。
arXiv:2608.12350v1 Announce Type: cross Abstract: The energy demand growth and environmental impacts of artificial intelligence (AI) have generated su…
3
支持深色模式：微软重塑 Win11 界面，下个目标锁定 AutoPlay 对话框
一键揪出应用卡顿根源，可视化分析性能数据，助力Win11优化提速。
IT之家 8 月 7 日消息，科技媒体 Windows Latest 今天（8 月 7 日）发布博文，报道称在 文件属性 、 运行 以及 打印管理工具 之后， 微软 WinUI 3 改造 Windows 11 界面的下个目标是自动播放（AutoPlay）对话框。 IT之家援引博文介绍，自动播放（Au…
4
揪出导致应用变慢根源，微软 Win11 性能分析工具测试接入 AI
接入AI的Win11性能分析工具，自动揪出CPU、磁盘、内存等卡顿根源，让性能调优不再靠手动翻数据。
IT之家 8 月 6 日消息，科技媒体 Windows Latest 昨日（8 月 5 日）发布博文，报道称微软正在测试在 Windows Performance Analyzer（WPA）接入 AI， 借助 MCP 让开发者分析 Windows 11 性能问题。 IT之家注：Windows Per…
5
DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis
DeepSeek V4 Flash 0731 在九大评测中的智能与性能表现如何？性价比是否值得入手？
Article URL: https://artificialanalysis.ai/models/deepseek-v4-flash Comments URL: https://news.ycombinator.com/item?id=49120299 Points: 369 # Comments…
6
Nothing Is Easy When You're an LLM: The Flat Latency Problem
大模型推理的“扁平延迟”困境，为何优化如此棘手？
Article URL: https://aidoses.substack.com/p/nothing-is-easy-when-youre-an-llm Comments URL: https://news.ycombinator.com/item?id=49125443 Points: 1 # …
7
Profiling Lightweight Large Language Models
揭秘轻量级大语言模型的内在工作原理与性能特征，为高效部署提供参考
arXiv:2607.20806v1 Announce Type: new Abstract: Lightweight large language models (LLMs) are increasingly being deployed locally on personal computers…
8
The Htop for LLM Inference
类似htop的LLM推理监控利器，一眼发现KV缓存瓶颈，还能深入分析运行负载。
Article URL: https://github.com/helasaoudi/llm-inspector Comments URL: https://news.ycombinator.com/item?id=48956776 Points: 1 # Comments: 0
9
wrk: كيفية اختبار تحميل API من سطر الأوامر
命令行下的 API 压测利器，轻松模拟高并发请求，秒级掌握服务性能极限
لقد نشرت نقطة نهاية API، وهي تعمل في المتصفح. لكن السؤال العملي هو: ماذا يحدث عندما يطلبها 400 مستخدم في الوقت نفسه؟ هل يبقى زمن الاستجابة مستقرًا؟ هل…
10
Show HN: AI harness for C/C++ with GDB, sanitizers, perf and compile tools
C/C++开发者福音，AI harness原生集成GDB、sanitizers等工具链，绕过常规AI编程助手的局限。
hey guys, i wanted to show one of my side projects. The idea is a coding harness (independent of models) natively designed for C/C++ developer workflo…
11
Optimizing Git Performance for Large Repositories
完整揭示Git大仓库性能优化路径，从跟踪耗时到精简克隆，手把手教你提升代码获取速度，是高性能开发必备指南。
Pinpointing Where Your Git Time Goes Squeezing Bytes: Packfile Tuning and Repository Cleanup Give Developers Only What They Need: Shallow, Sparse, and…
12
Little Brains, Big Feats: Exploring Compact Language Models
紧凑模型也能有大本事，看小参数如何撬动大语言模型性能，数据科学顶会研究值得一读。
arXiv:2606.30062v1 Announce Type: new Abstract: While large language models have been dominating the research landscape recently, small language model…
13
SOLAR: AI-Powered Speed-of-Light Performance Analysis
用AI实现光速级性能分析，突破传统瓶颈，硬件优化新范式。
arXiv:2606.26383v1 Announce Type: cross Abstract: How fast could a deep-learning model run on target hardware, and how far is today's implementation f…
14
Ask HN: Has Codex gotten slower recently?
开发者吐槽Codex变慢？真实性能波动调查，用AI写代码前必看
Comments URL: https://news.ycombinator.com/item?id=48639619 Points: 6 # Comments: 1
15
What 'Getting Your Hands Dirty' Means at LLM-Era
LLM时代，真正的“脏活”已从写CRUD转向排查生产故障，重新定义工程师的核心价值。
Article URL: https://carette.xyz/posts/the_mud_and_the_mind/ Comments URL: https://news.ycombinator.com/item?id=48596273 Points: 8 # Comments: 1
16
Moving beyond fork() + exec()
一篇深度解析Linux内核进程创建优化技术的文章，附代码补丁细节与性能分析
Article URL: https://lwn.net/SubscriberLink/1076018/16f01bbbb8e0d1f0/ Comments URL: https://news.ycombinator.com/item?id=48425528 Points: 257 # Commen…
17
Show HN: Mercek – A Desktop IDE for AWS ECS
一款专为AWS ECS打造的桌面IDE，集请求监控、成本分析、配置读取于一体，让云服务管理更高效。
Hey HN I've been using ECS for a while now and found it annoying having to log into the console everytime I use Lens for Kubernetes but couldnt find a…
18
Don't Gamble, GAMBLe: An Analytical Framework for AI-Driven Research Systems
提出GAMBLe分析框架，为AI驱动研究系统提供系统性评估方法，避免盲目探索。
arXiv:2606.02863v1 Announce Type: new Abstract: AI-Driven Research Systems (ADRS) -- systems coupling LLMs with automated evaluation to discover algor…
19
Memory-Bound but Not Bandwidth-Limited: The Physical AI Inference Gap in Batch-1 LLM Decode
揭示LLM推理瓶颈新视角：batch-1解码受内存限制而非带宽限制，挑战传统认知。
arXiv:2605.30571v1 Announce Type: cross Abstract: Physical AI systems, including robots, autonomous vehicles, embodied agents and edge copilots, often…
20
Show HN: Live, system-wide USB transfer sniffer in eBPF
基于eBPF的系统级USB流量嗅探器，支持请求/响应配对、控制传输解码与延迟直方图，调试USB设备的利器。
Article URL: https://github.com/yeet-src/usbsnoop Comments URL: https://news.ycombinator.com/item?id=48342821 Points: 4 # Comments: 0