#
ENJA

local-llm

jamesob/local-llm

A comprehensive hardware and software configuration guide for building and operating a high-performance local LLM and speech-to-text inference cluster.

1.8k 107No licenseUpdated 2026-07-10

Overview

This repository provides a detailed blueprint and runner configurations for setting up state-of-the-art local AI infrastructure, featuring a custom 4x GPU rig with PCIe peer-to-peer switching, Docker-based model serving, and integrated local tooling for agentic workflows.

Capabilities

  • Hardware bill of materials for local AI rigs at various price points
  • PCIe switch configuration, BIOS bifurcation, and kernel parameter tuning for multi-GPU setups
  • Ready-to-run Docker Compose serving configurations for large models like GLM and Qwen
  • Local speech-to-text pipeline setup using cohere-transcribe
  • Custom scripts for measuring GPU peer-to-peer bandwidth and latency

Best for

Building high-performance local GPU clusters for running state-of-the-art large language models, Configuring low-latency peer-to-peer GPU interconnects using custom PCIe switches, Deploying local speech-to-text transcription harnesses, Setting up containerized model serving runners with Docker Compose

Works with

OpenCode
Evoa Score breakdown= Σ (score × weight)
61
Task usefulness20%88 → +17.6

実タスクにどれだけ役立つか(機能の豊富さ・用途の明確さ)。 AIによるcapabilities/use-cases解析

Code quality15%85 → +12.8

実装・指示の品質。 AIによるSKILL.md/README解析

Maintenance15%85 → +12.8

リポジトリがどれだけ活発に保守されているか。 GitHub 最終push日時の新しさ

Documentation12%33 → +4.0

ドキュメントの充実度・分かりやすさ。 README/独自要約の情報量

Security15%50 → +7.5

危険・不審な挙動が無いか。 AIによるセキュリティレビュー

Originality10%95 → +9.5

ありふれたラッパーではない独自性。 AIによる独自性判定

Popularity8%82 → +6.6

コミュニティの採用度。 GitHub Stars/Forks(対数スケール)

Compatibility5%25 → +1.3

対応AIエージェントの広さ。 AIによる対応エージェント判定

Weighted total71.9 / 100

ライセンス不明/制限あり(Red)のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →

Security considerations

Instructions include disabling IOMMU and PCIe ACS at the kernel/runtime level to enable peer-to-peer GPU traffic, which alters standard system security isolation boundaries.

Categories

Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.