JoinGrove

Adaptive Performance Profiles

Source: docs/global-south/performance.md
Generated: 01/09/2026

JoinGrove auto-detects RAM and CPU on startup and applies lite, standard, or full performance profiles.

Key Concepts

  • What: Adaptive performance tunes UI motion, editor features, default Ollama model tier, and build recommendations for the host hardware.
  • Why: 8 GB laptops and shared lab VMs are common in Global South bootcamps — defaults must not assume 32 GB MacBook Pros.
  • Related: Ollama tiers (offline-providers.md), model routing (routing.md), run anywhere (run-modes.md).

API / Reference

Profile Typical hardware Defaults applied
Lite ≤ 8 GB RAM, low CPU Reduced motion, no minimap, qwen2.5-coder:1.5b, build-fast hint
Standard 16 GB RAM Balanced UI, qwen2.5-coder:7b
Full High-spec workstation All features enabled, user model choice

Related settings

Setting / feature Purpose
joingroveRunAnywhere.liteBuildCommand npm run build-fast for campus VMs
Low-spec defaults Manual updates off, telemetry off
Adaptive Ollama pull Setup script recommends model by RAM

Setup & Installation

  1. JoinGrove detects hardware on startup — no manual config required.
  2. For campus VMs → use npm run build-fast instead of full compile.
  3. Ollama setup → npm run setup-offline-ai picks lite vs standard model guidance.
  4. Command Palette → Choose How to Run JoinGrove… for REH on low-spec server.

Code Examples

Fast build for campus server:

npm run build-fast

Setup offline AI (adaptive model hints):

npm run setup-offline-ai

Smoke test run-anywhere + Ollama:

npm run smoke-run-anywhere -- --check-ollama

Next: See offline-providers.md for RAM → model mapping table.

How It Connects