treeru.com

TreeRU Tech Blog

Insights and hands-on guides on frontend, backend, databases, IT infrastructure, and AI solutions.

Trending Top 5
Network2026-08-04

Our Servers Catch Burglars Every Day — TREEWALL Field Test Report

Hardware2026-07-05

NVMe Storage Benchmark: Optane to HDD — 20 Devices Compared with fio

AI2026-07-02

Local AI Video Generation with ComfyUI + Wan 2.2 — An RTX PRO 6000 96GB Build Log

AI2026-07-02

Structuring Unstructured Documents with an LLM — Measured Accuracy on a 19-Item Menu

Network2026-06-13

Guarding the Front Door of Your Network — TREEWALL, an AI Security Gateway

posts.tsx — 28 posts
Load More18 remaining
  1. Local LLM Benchmark: 6 Models Tested Across 60 Questions and 7 Business Scenarios
  2. Building a Local RAG Pipeline — From Embedding Model Selection to Hallucination Elimination
  3. MoE vs Dense: Why Qwen3-30B-A3B Is Slower Than 14B — And Why Hybrid Inference Failed
  4. Qwen3-32B vs 14B — Is 2x Slower Speed Worth the Quality Gain?
  5. Text2SQL Real-World Test — When LLM Writes SQL Directly
  6. AWQ Quantization Speed Benchmark: 16 Models, INT4 vs BF16, and the MoE Reversal
  7. SGLang 23-Model Serving Guide — Optimal Configuration for Every Model
  8. Qwen3-14B Deep Review — Why It Is Our Top-Ranked Local LLM
  9. Cross-Server AI Inference — Boosting Throughput 70% with a $450 Secondary GPU
  10. Serving 7 Companies on 1 GPU — Multi-Tenant Isolation Testing in Practice
  11. LoRA Fine-Tuning for Custom AI Chatbots — From 10 Training Pairs to Multi-Tenant Serving
  12. SGLang vs vLLM — The Secret Behind the 3x Throughput Gap
  13. Local LLM Concurrent User Load Test — How Many Users Can an RTX PRO 6000 Handle?
  14. Local LLM Business Test (Part 2) — Shopping, Legal & Automation
  15. Local LLM Business Test (Part 1) — Manufacturing, SaaS, Healthcare
  16. LLM Hallucination Test — Which Local Models Fabricate Information?
  17. Local LLM Korean Language Comparison — 6 Models Tested with 10 Real Questions
  18. 8B vs 14B vs 32B LLM: Concurrent User Benchmark on a Single GPU
Editor's Picks
Network

Build a CCTV Recording Server Without an NVR — RTSP + ffmpeg + systemd, 180-Day Retention

AI

Our MoE Model Started Spewing Garbage — What Auditing All 18,432 Expert Weights Revealed

AI

What We Learned Building and Auditing 990 AI-Generated English Learning Scenes

AI

BGE-M3 vs mE5-Large — Comparing Korean RAG Search Quality Head to Head

Tools

Recovering winget and Windows Terminal on a Windows 11 Box Where the Microsoft Store Won't Open