Update dependencies for PyTorch, for system cuda13

README.md

@@ -33,179 +33,17 @@ This repository includes external repositories as Git submodules, so, unless you
 git submodule update --init --recursive
 ```
 
-### Create a Conda Environment
+GTSfM supports two installation methods. Choose the one that best fits your workflow:
 
-To run GTSfM, first, we need to create a conda environment with the required dependencies.
+### Option 1: Conda Setup (Recommended for most users)
 
-[Install MiniConda](https://www.anaconda.com/docs/getting-started/miniconda/install) if needed, then:
+For detailed Conda installation instructions, see **[conda-setup.md](docs/setup/conda-setup.md)**
 
-On **Linux**, with CUDA support, run:
-```bash
-conda env create -f environment_linux.yml
-conda activate gtsfm-v1 # you may need "source activate gtsfm-v1" depending upon your bash and conda set-up
-```
-Check your cuda version then install `torch_scatter` from pre-built wheels
-
-For example, for CUDA 12.1 → use cu121 
-```bash
-pip install torch-scatter -f https://data.pyg.org/whl/torch-2.5.1+cu121.html
-```
-
-To use PACE, use the same linux installation `environment_linux.yml`
-
-Then add `dask-cuda`
-```bash
-conda install -c rapidsai -c conda-forge dask-cuda
-```
-
-
-On **macOS**, there is no CUDA support, so run:
-
-```bash
-conda env create -f environment_mac.yml
-conda activate gtsfm-v1
-```
-
-### Install `gtsfm` as a module
-
-Now, install `gtsfm` as a module:
-
-```bash
-pip install -e .
-```
-
-Make sure that you can run `python -c "import gtsfm; import gtsam; print('hello world')"` in python, and you are good to go!
-
-
-#### Set Up GTSfM with UV
-
-**Basic Installation:**
-Navigate to the GTSfM directory:
-
-```bash
-cd path/to/gtsfm
-
-# Clean existing environment (if any)
-rm -rf .venv
-
-# Install on Linux for CPU only and macOS
-uv sync --python 3.10
-
-# Install on Linux with CUDA GPU
-uv sync --python 3.10 --extra complete
-
-# Install torch-scatter (platform-specific)
-# For Linux (CUDA 12.1):
-uv pip install torch-scatter --find-links https://data.pyg.org/whl/torch-2.5.1+cu121.html
-
-```
-
-**Multi-GPU Installation (For Distributed Computing):**
-
-If you have multiple GPUs on the same machine and want to use Dask for distributed GPU computing:
-
-```bash
-# Multiple GPUs per node (e.g., 4x or 8x A100)
-uv sync --python 3.10 --extra complete --extra multi-gpu
-```
-
-This adds `dask-cuda` for GPU-aware distributed scheduling.
-
-**When do you need `--extra multi-gpu`?**
-- You have multiple GPUs on the same machine
-- You want to use Dask to distribute work across GPUs
-- You're running on a GPU cluster node
-
-**When you DON'T need it:**
-- Single GPU workstation
-- Laptop with one GPU
-- CPU-only machines
-- Multiple machines (handled differently)
+### Option 2: UV Setup (Fast alternative package manager)
 
-**Add System Level Package**
-```bash
-# Linux
-sudo apt-get install nodejs npm
-sudo apt-get install graphviz
-# macOS
-brew install node
-brew install graphviz
-```
-
-**Verify installation**
-```bash
-uv run python -c "import gtsfm; import pydegensac; import torch; import torch_scatter; print('✅ Success!')"
-```
-
-#### Go to test using UV
-```bash
-uv run ./run --dataset_dir tests/data/set1_lund_door \
---config_name unified_binary.yaml \
---loader olsson \
---num_workers 2 graph_partitioner.max_depth=1
-```
-
-```bash
-uv run .github/scripts/execute_single_benchmark.sh skydio-8 lightglue 15 colmap-loader 760 true
-```
-
-#### Managing Packages with UV
-
-**Adding a new package:**
-```bash
-uv add <package-name>
-# Example: uv add numpy
-```
-
-**Adding a development dependency:**
-```bash
-uv add --dev <package-name>
-# Example: uv add --dev pytest
-```
-
-**Removing a package:**
-```bash
-uv remove <package-name>
-# Example: uv remove numpy
-```
-
-**Installing a package without adding to dependencies:**
-```bash
-uv pip install <package-name>
-```
-
-#### When to Use `uv lock`
-
-The `uv lock` command updates the lock file (`uv.lock`) without installing packages. Use it when:
-
-- **After manually editing `pyproject.toml`:** When you directly modify dependencies in the configuration file
-- **To update dependencies:** When you want to resolve and lock new versions without installing
-- **In CI/CD pipelines:** To ensure reproducible builds by generating a lock file
-- **Before committing changes:** To update the lock file for team members
-
-```bash
-# Update lock file after editing pyproject.toml
-uv lock
-
-# Update lock file and install packages
-uv lock && uv sync
-```
-
-> **Note:** `uv add` and `uv remove` automatically update the lock file, so you typically don't need to run `uv lock` manually after these commands.
-
-#### Running Commands with UV
-
-Once set up, prefix Python commands with `uv run`:
-
-```bash
-# Run GTSfM
-uv run python -m gtsfm.runner --config_name deep_front_end.yaml
-
-# Or use the run script
-uv run ./run --dataset_dir tests/data/set1_lund_door --config_name deep_front_end.yaml
-```
+For detailed UV installation instructions, see **[uv-setup.md](docs/setup/uv-setup.md)**
 
-Make sure that you can run `python -c "import gtsfm; import gtsam; print('hello world')"` in python, and you are good to go!
+Both methods will allow you to run GTSfM successfully.
 
 ## Try It on Google Colab  
 
@@ -320,7 +158,7 @@ Refer to the detailed guide:
 ### Running GTSFM on a Multi-Machine Cluster  
 
 For users who want to run GTSFM on a **cluster of multiple machines**, follow the setup instructions here:  
-📄 [CLUSTER.md](https://github.com/borglab/gtsfm/tree/master/CLUSTER.md)  
+📄 [CLUSTER.md](https://github.com/borglab/gtsfm/tree/master/docs/deployment/CLUSTER.md)  
 
 ### Where Are the Results Stored?  
 
@@ -422,7 +260,7 @@ GTSfM is designed in a modular way. Each module can be swapped out with a new on
 
 ## Contributing
 
-Contributions are always welcome! Please be aware of our [contribution guidelines for this project](CONTRIBUTING.md).
+Contributions are always welcome! Please be aware of our [contribution guidelines for this project](docs/CONTRIBUTING.md).
 
 ## Citing this work
 

docs/setup/conda-setup.md

@@ -0,0 +1,51 @@
+# Conda Setup Guide
+
+This guide covers installation of GTSfM using Conda package manager.
+
+## Prerequisites
+
+- [MiniConda](https://www.anaconda.com/docs/getting-started/miniconda/install) installed on your system
+
+## Create a Conda Environment
+
+To run GTSfM, first, we need to create a conda environment with the required dependencies.
+
+### Linux (with CUDA support)
+
+```bash
+conda env create -f environment_linux.yml
+conda activate gtsfm-v1 # you may need "source activate gtsfm-v1" depending upon your bash and conda set-up
+```
+
+Check your cuda version then install `torch_scatter` from pre-built wheels
+
+For example, for CUDA 12.1, use cu121 
+```bash
+pip install torch-scatter -f https://data.pyg.org/whl/torch-2.5.1+cu121.html
+```
+
+For CUDA 13, use cu128
+```bash
+pip install torch-scatter --find-links https://data.pyg.org/whl/torch-2.7.0+cu128.html
+```
+
+### macOS (no CUDA support)
+
+```bash
+conda env create -f environment_mac.yml
+conda activate gtsfm-v1
+```
+
+## Install `gtsfm` as a module
+
+Now, install `gtsfm` as a module:
+
+```bash
+pip install -e .
+```
+
+## Verify Installation
+
+
+Make sure that you can run `python -c "import gtsfm; import gtsam; print('hello world')"` in python, and you are good to go!
+