GPU Monitor for AI Workloads v1.3.0

Changes

• Tray-only minimize: App now hides from taskbar when minimized, staying only in system tray
• Linux build: Added AppImage and tar.gz targets
• Build fix: Updated libasound2 → libasound2t64 for Ubuntu 24.04 compatibility

GPU Monitor for AI Workloads v1.2.0

Changes

• GPU Name Resolution: Fixed GPU naming collision using DEVICE_ID/SUBSYSTEM_ID lookup table
  • GPU 0: AMD Radeon AI PRO R9700 (32GB)
  • GPU 1: Radeon RX 9060 XT (16GB)
  • GPU 2: iGPU 512MB (auto-detected from Ryzen CPU)
• NSIS Build Fix: Removed incompatible SetWindowsVersion command
• Unified Monitor Agent: Added Intel XPU support alongside AMD ROCm

Artifacts

• Installer: GPU-Monitor-AI-1.2.0-win64-installer.exe (79MB)
• Portable: GPU-Monitor-AI-1.2.0-win64-portable.zip (108MB)

GPU Monitor for AI Workloads v1.1.0

What's New in v1.1.0

Settings Modal Fixes

• Wider modal (560px) for better breathing room and readability
• Source names now display fully — no more truncation
• Reorder handle (⠇) renders inline with the source name
• Info tooltip switched from hover to click-toggle — more reliable in Electron's Chromium
• Inline “Test Connection” button added next to the info icon

Build System

• Migrated electron-builder config to separate electron-builder.yml for v25 compatibility
• Added portable zip build alongside NSIS installer

Installation

Option 1 — Portable (Recommended):
Download GPU-Monitor-AI-1.1.0-win64-portable.zip, extract, and run GPU Monitor for AI Workloads.exe directly.

Option 2 — Installer:
Download GPU-Monitor-AI-1.1.0-win64-installer.exe for a traditional installer with desktop shortcuts.

Full Changelog

See commits on master.

GPU Monitor for AI Workloads v1.0.0

What's New

First stable release of GPU Monitor for AI Workloads.

Features

• Unified Dashboard — AMD ROCm, NVIDIA, and Intel XPU GPUs in one window
• Draggable Tiles — reorder cards to prioritise what matters most
• Source Filtering — filter by vendor with one click
• Dark Navy Theme — easy on the eyes during long training sessions
• Custom Window Chrome — frameless window with native minimise/maximise/close controls
• Configurable Sources — add/remove HTTP sources via Settings
• Auto-Polling — configurable refresh interval (default: 2 seconds)
• Always-on-Top Mode — keep the dashboard visible above your IDE

Supported GPUs

Vendor	Detection Method	Platform
NVIDIA	nvidia-smi (local)	Windows
AMD ROCm	HTTP Agent (remote Linux)	Windows + Linux
Intel XPU	HTTP Agent (remote Linux)	Windows + Linux

What's Included

• Installer — NSIS installer with desktop shortcut and Start Menu entry
• Unpacked — portable version for side-by-side testing
• Source Code — Electron app + Flask agents (agent/ directory)

Quick Start

1. Run the installer
2. Launch GPU Monitor for AI Workloads from the desktop shortcut
3. Add HTTP sources in Settings -> Sources tab (for AMD/Intel GPUs)
4. Local NVIDIA GPUs are detected automatically

Full documentation