This repo contains clean, tested, and practical guides for setting up GPU environments on Linux servers, including:
- β Clean install of NVIDIA drivers + CUDA Toolkit 12.6
- π Compatibility handling for RTX 2xxx and 3xxx series GPUs
- π Future: Apex, TensorBoard, PyTorch setup, multi-GPU tips
Guide | Description |
---|---|
CUDA_12.6_Installation_Guide.md | Step-by-step clean install for CUDA 12.6 and driver setup |
(Coming Soon) Apex Setup Guide | How to install NVIDIA Apex for mixed-precision training |
(Coming Soon) TensorBoard + PyTorch | How to log and visualize training easily |
- Solves real pain points in driver mismatch,
nvcc
not found,nvidia-smi
broken - Focused on minimal + working setup for real-world ML/DL projects
- Easy to copy-paste, no fluff
- Ubuntu 20.04 / 22.04
- NVIDIA RTX 2xxx, 3xxx, 4xxx series
- Tested on Conda-based Python environments
Found a better way or a newer driver? PRs and issues welcome!