mirror of
https://github.com/forkless/XPUSYS-Monitor-NG.git
synced 2026-08-16 00:46:37 +02:00
docs: add README.md for XPUSYS-Monitor-NG
This commit is contained in:
@@ -0,0 +1,66 @@
|
||||
# XPUSYS-Monitor-NG
|
||||
|
||||
> **Proof of Concept** — Native Windows ROCm support for ComfyUI-XPUSYS-Monitor
|
||||
|
||||
This is a fork of [ComfyUI-XPUSYS-Monitor](https://github.com/allanmeng/ComfyUI-XPUSYS-Monitor)
|
||||
by allanmeng, created as a proof of concept to add **native Windows AMD ROCm
|
||||
support** without requiring `rocm_smi_lib` (a Linux-only Python package).
|
||||
|
||||
## Intent
|
||||
|
||||
The upstream project is an excellent hardware monitor for ComfyUI, but its AMD
|
||||
GPU support depended on `rocm_smi_lib`, which cannot be installed on Windows.
|
||||
This POC replaces the AMD provider's GPU monitoring with Windows-native APIs:
|
||||
|
||||
- **VRAM**: `torch.cuda.mem_get_info()` — works on ROCm 6+ for Windows
|
||||
- **GPU load**: `typeperf` (Windows built-in) — reads WDDM engine counters
|
||||
- **Temperature / Power / Clock**: Unavailable — the AMD Windows driver does
|
||||
not expose these through standard Python-accessible APIs on this hardware
|
||||
|
||||
## Relationship to Upstream
|
||||
|
||||
This project has **no intention to fork** the original. The changes here are
|
||||
narrowly scoped to making the AMD provider work on Windows ROCm. AI agent-
|
||||
assisted modifications were used during development, and the author does not
|
||||
want to push AI-generated changes into allanmeng's original repository.
|
||||
|
||||
If the upstream maintainer wishes to incorporate any of these changes, they
|
||||
are welcome to take whatever is useful.
|
||||
|
||||
## Status
|
||||
|
||||
| Metric | Source | Status |
|
||||
|---|---|---|
|
||||
| VRAM free / total | `torch.cuda.mem_get_info(0)` | ✅ |
|
||||
| VRAM allocated / reserved | `torch.cuda.memory_allocated()` / `memory_reserved()` | ✅ |
|
||||
| GPU utilisation | `typeperf` WDDM engine counters (max) | ✅ |
|
||||
| Device name | `torch.cuda.get_device_name(0)` | ✅ |
|
||||
| GPU core temperature | Driver does not expose via Windows API | ❌ |
|
||||
| Power draw | Driver does not expose via Windows API | ❌ |
|
||||
| Core clock | Driver does not expose via Windows API | ❌ |
|
||||
| `rocm_smi_lib` dependency | Removed; no replacement needed | ✅ |
|
||||
| Windows ROCm detection | `getattr(torch.version, 'roc', None)` + GPU name fallback | ✅ |
|
||||
|
||||
## Requirements
|
||||
|
||||
- ComfyUI (any recent version)
|
||||
- [ROCm for Windows](https://rocm.docs.amd.com/) 6+ (tested on ROCm 7.2)
|
||||
- PyTorch ROCm build (e.g. `pytorch 2.9.1+rocm7.2.1`)
|
||||
- `psutil` (`pip install psutil`)
|
||||
|
||||
## Installation
|
||||
|
||||
Clone into ComfyUI `custom_nodes/`:
|
||||
|
||||
```cmd
|
||||
cd ComfyUI/custom_nodes
|
||||
git clone https://github.com/forkless/XPUSYS-Monitor-NG
|
||||
cd XPUSYS-Monitor-NG
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
|
||||
Restart ComfyUI. The plugin auto-detects AMD ROCm and loads the AMDProvider.
|
||||
|
||||
## License
|
||||
|
||||
Same as upstream — MIT.
|
||||
Reference in New Issue
Block a user