sarman/dgx-spark-playbooks
1
0
Fork 0
mirror of https://github.com/NVIDIA/dgx-spark-playbooks.git synced 2026-04-22 18:13:52 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

chore: Regenerate all playbooks

Browse Source
This commit is contained in:
GitLab CI 2025-10-05 18:20:38 +00:00
parent 53f06ed06c
commit f39dd60161
5 changed files with 68 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@ -25,7 +25,7 @@ Each playbook includes prerequisites, step-by-step instructions, troubleshooting
- [Connect to your Spark](nvidia/connect-to-your-spark/)
- [DGX Dashboard](nvidia/dgx-dashboard/)
- [FLUX.1 Dreambooth LoRA Fine-tuning](nvidia/flux-finetuning/)
- [Optimized Jax](nvidia/jax/)
- [Optimized JAX](nvidia/jax/)
- [Llama Factory](nvidia/llama-factory/)
- [MONAI-Reasoning-CXR-3B Model](nvidia/monai-reasoning/)
- [Build and Deploy a Multi-Agent Chatbot](nvidia/multi-agent-chatbot/)

30
nvidia/jax/README.md
View File

@ -1,6 +1,6 @@
# Optimized Jax
# Optimized JAX
> Develop with Optimized Jax
> Develop with Optimized JAX
## Table of Contents
@ -17,10 +17,10 @@ JAX lets you write **NumPy-style Python code** and run it fast on GPUs without w
- **NumPy on accelerators**: Use `jax.numpy` just like NumPy, but arrays live on the GPU.
- **Function transformations**:
- `jit` → Compiles your function into fast GPU code.
- `grad` → Gives you automatic differentiation.
- `vmap` → Vectorizes your function across batches.
- `pmap` → Runs across multiple GPUs in parallel.
- `jit` → Compiles your function into fast GPU code
- `grad` → Gives you automatic differentiation
- `vmap` → Vectorizes your function across batches
- `pmap` → Runs across multiple GPUs in parallel
- **XLA backend**: JAX hands your code to XLA (Accelerated Linear Algebra compiler), which fuses operations and generates optimized GPU kernels.
## What you'll accomplish
@ -40,12 +40,12 @@ GPU acceleration and performance optimization capabilities.
## Prerequisites
[ ] NVIDIA Spark device with Blackwell architecture
[ ] ARM64 (AArch64) processor architecture
[ ] Docker or container runtime installed
[ ] NVIDIA Container Toolkit configured
[ ] Verify GPU access: `nvidia-smi`
[ ] Port 8080 available for marimo notebook access
- NVIDIA Spark device with Blackwell architecture
- ARM64 (AArch64) processor architecture
- Docker or container runtime installed
- NVIDIA Container Toolkit configured
- Verify GPU access: `nvidia-smi`
- Port 8080 available for marimo notebook access
## Ancillary files
@ -119,7 +119,7 @@ docker run --gpus all --rm -it \
jax-on-spark
```
## Step 5. Access marimo interface
## Step 5. Access the marimo interface
Connect to the marimo notebook server to begin the JAX tutorial.
@ -130,7 +130,7 @@ Connect to the marimo notebook server to begin the JAX tutorial.
The interface will load a table-of-contents display and brief introduction to marimo.
## Step 6. Complete JAX introduction tutorial
## Step 6. Complete the JAX introduction tutorial
Work through the introductory material to understand JAX programming model differences from NumPy.
@ -172,7 +172,7 @@ Common issues and their solutions:
| Symptom | Cause | Fix |
|---------|--------|-----|
| `nvidia-smi` not found | Missing NVIDIA drivers | Install NVIDIA drivers for ARM64 |
| Container fails to access GPU | Missing NVIDIA Container Toolkit | Install nvidia-container-toolkit |
| Container fails to access GPU | Missing NVIDIA Container Toolkit | Install `nvidia-container-toolkit` |
| JAX only uses CPU | CUDA/JAX version mismatch | Reinstall JAX with CUDA support |
| Port 8080 unavailable | Port already in use | Use `-p 8081:8080` or kill process on 8080 |
| Package conflicts in Docker build | Outdated environment file | Update environment file for Blackwell |

2
nvidia/nccl/README.md
View File

@ -13,7 +13,7 @@
## Overview
## Basic Idea
## Basic idea
NCCL (NVIDIA Collective Communication Library) enables high-performance GPU-to-GPU communication
across multiple nodes. This walkthrough sets up NCCL for multi-node distributed training on

22
nvidia/rag-ai-workbench/README.md
View File

@ -14,7 +14,7 @@
## Overview
## Basic Idea
## Basic idea
This walkthrough demonstrates how to set up and run an agentic retrieval-augmented generation (RAG)
project using NVIDIA AI Workbench. You'll use AI Workbench to clone and run a pre-built agentic RAG
@ -39,16 +39,16 @@ architectures.
## Prerequisites
- [ ] DGX Spark system with NVIDIA AI Workbench installed or ready to install
- [ ] Free NVIDIA API key: Generate at [NGC API Keys](https://org.ngc.nvidia.com/setup/api-keys)
- [ ] Free Tavily API key: Generate at [Tavily](https://tavily.com/)
- [ ] Internet connection for cloning repositories and accessing APIs
- [ ] Web browser for accessing the Gradio interface
- DGX Spark system with NVIDIA AI Workbench installed or ready to install
- Free NVIDIA API key: Generate at [NGC API Keys](https://org.ngc.nvidia.com/setup/api-keys)
- Free Tavily API key: Generate at [Tavily](https://tavily.com/)
- Internet connection for cloning repositories and accessing APIs
- Web browser for accessing the Gradio interface
**Verification commands:**
## Verification commands
* Verify the NVIDIA AI Workbench application exists on your DGX Spark system
* Verify your API keys are valid and up-to-date
- Verify the NVIDIA AI Workbench application exists on your DGX Spark system
- Verify your API keys are valid and up-to-date
## Time & risk
@ -74,7 +74,7 @@ On your DGX Spark system, open the **NVIDIA AI Workbench** application and click
### Troubleshooting installation issues
If you encounter the error message: `An error occurred ... container tool failed to reach ready state. try again: docker is not running`, reboot your DGX Spark system to restart the docker service, then reopen NVIDIA AI Workbench.
If you encounter the error message: `An error occurred ... container tool failed to reach ready state. try again: docker is not running` reboot your DGX Spark system to restart the docker service, then reopen NVIDIA AI Workbench.
## Step 2. Verify API key requirements
@ -94,7 +94,7 @@ This step clones the pre-built agentic RAG project from GitHub into your AI Work
From the AI Workbench landing page, select the **Local** location if not done so already, then click **Clone Project** from the top right corner.
Paste this Git repository URL in the clone dialog: ``https://github.com/NVIDIA/workbench-example-agentic-rag``.
Paste this Git repository URL in the clone dialog: https://github.com/NVIDIA/workbench-example-agentic-rag
Click **Clone** to begin the clone and build process.

75
nvidia/vscode/README.md
View File

@ -12,8 +12,8 @@
## Overview
## Basic Idea
This walkthrough establishes a local Visual Studio Code development environment directly on DGX Spark devices. By installing VSCode natively on the ARM64-based Spark system, you gain access to a full-featured IDE with extensions, integrated terminal, and Git integration while leveraging the specialized hardware for development and testing.
## Basic idea
This walkthrough establishes a local Visual Studio Code development environment directly on DGX Spark devices. By installing VS Code natively on the ARM64-based Spark system, you gain access to a full-featured IDE with extensions, integrated terminal, and Git integration while leveraging the specialized hardware for development and testing.
## What you'll accomplish
You will have Visual Studio Code running natively on your DGX Spark device with access to the system's ARM64 architecture and GPU resources. This setup enables direct code development, debugging, and execution on the target hardware without remote development overhead.
@ -30,7 +30,7 @@ You will have Visual Studio Code running natively on your DGX Spark device with
• DGX Spark device with administrative privileges
• Active internet connection for downloading the VSCode installer
• Active internet connection for downloading the VS Code installer
• Verify ARM64 architecture:
```bash
@ -56,22 +56,22 @@ You will have Visual Studio Code running natively on your DGX Spark device with
## Step 1. Verify system requirements
Before installing VSCode, confirm your DGX Spark system meets the requirements and has GUI support.
Before installing VS Code, confirm your DGX Spark system meets the requirements and has GUI support.
```bash
## Verify ARM64 architecture
uname -m
## Check available disk space (VSCode requires ~200MB)
## Check available disk space (VS Code requires ~200MB)
df -h /
## Verify desktop environment is running
ps aux | grep -E "(gnome|kde|xfce)"
```
## Step 2. Download VSCode ARM64 installer
## Step 2. Download VS Code ARM64 installer
Navigate to the VSCode [download](https://code.visualstudio.com/download) page and download the appropriate ARM64 `.deb` package for your system.
Navigate to the VS Code [download](https://code.visualstudio.com/download) page and download the appropriate ARM64 `.deb` package for your system.
Alternatively, you can download the installer with this command:
@ -79,7 +79,7 @@ Alternatively, you can download the installer with this command:
wget https://code.visualstudio.com/sha/download?build=stable\&os=linux-deb-arm64 -O vscode-arm64.deb
```
## Step 3. Install VSCode package
## Step 3. Install VS Code package
Install the downloaded package using the system package manager.
@ -95,29 +95,29 @@ sudo apt-get install -f
## Step 4. Verify installation
Confirm the VSCode app is installed successfully and can launch.
Confirm the VS Code app is installed successfully and can launch.
You can open the app directly from the list of applications or use the command line.
```bash
## Check if VSCode is installed
## Check if VS Code is installed
which code
## Verify version
code --version
## Test launch (will open VSCode GUI)
## Test launch (will open VS Code GUI)
code &
```
VSCode should launch and display the welcome screen.
VS Code should launch and display the welcome screen.
## Step 5. Configure for Spark development
Set up VSCode for development on the DGX Spark platform.
Set up VS Code for development on the DGX Spark platform.
```bash
## Launch VSCode if not already running
## Launch VS Code if not already running
code
## Or create a new project directory and open it
@ -126,7 +126,7 @@ cd ~/spark-dev-workspace
code .
```
From within VSCode:
From within VS Code:
* Open **File** > **Preferences** > **Settings**
* Search for "terminal integrated shell" to configure default terminal
@ -134,7 +134,7 @@ From within VSCode:
## Step 6. Validate setup and test functionality
Test core VSCode functionality to ensure proper operation on ARM64.
Test core VS Code functionality to ensure proper operation on ARM64.
Create a test file:
```bash
@ -145,7 +145,7 @@ echo 'print("Hello from DGX Spark!")' > test.py
code test.py
```
Within VSCode:
Within VS Code:
* Verify syntax highlighting works
* Open integrated terminal (**Terminal** > **New Terminal**)
* Run the test script: `python3 test.py`
@ -156,16 +156,16 @@ Within VSCode:
| Symptom | Cause | Fix |
|---------|-------|-----|
| `dpkg: dependency problems` during install | Missing dependencies | Run `sudo apt-get install -f` |
| VSCode won't launch with GUI error | No display server/X11 | Verify GUI desktop is running: `echo $DISPLAY` |
| VS Code won't launch with GUI error | No display server/X11 | Verify GUI desktop is running: `echo $DISPLAY` |
| Extensions fail to install | Network connectivity or ARM64 compatibility | Check internet connection, verify extension ARM64 support |
## Step 8. Uninstalling VSCode
## Step 8. Uninstalling VS Code
> **Warning:** Uninstalling VSCode will remove all user settings and extensions.
> **Warning:** Uninstalling VS Code will remove all user settings and extensions.
To remove VSCode if needed:
To remove VS Code if needed:
```bash
## Remove VSCode package
## Remove VS Code package
sudo apt-get remove code
## Remove configuration files (optional)
@ -175,23 +175,28 @@ rm -rf ~/.vscode
## Access with NVIDIA Sync
## Step 1. Install and Open NVIDIA Sync
## Step 1. Install and configure NVIDIA Sync
## Step 2. Add your Spark to NVIDIA Sync
Follow the [NVIDIA Sync setup guide](/spark/connect-to-your-spark/sync) to:
- Install NVIDIA Sync for your operating system
- Configure which development tools you want to use (VS Code, Cursor, Terminal, etc.)
- Add your DGX Spark device by providing its hostname/IP and credentials
## Step 3. Install VS Code locally
NVIDIA Sync will automatically configure SSH key-based authentication for secure, password-free access.
## Step 4. Open Sync and launch VS Code
## Step 2. Launch VS Code through NVIDIA Sync
- Click the NVIDIA Sync icon in your system tray/taskbar
- Ensure your device is connected (click "Connect" if needed)
- Click on "VS Code" to launch it with an automatic SSH connection to your Spark
- Wait for the remote connection to be established (may ask your local machine for a password or to authorize the connection)
- It may prompt you to "trust the authors of the files in this folder" when you first land in the home directory after a successful ssh connection.
- It may prompt you to "trust the authors of the files in this folder" when you first land in the home directory after a successful SSH connection
## Step 3. Validation and follow-ups
## Step 5. Validation and Follow-ups
- Verify that you can access your Spark's filesystem with VSCode as a text editor. Run test commands in the terminal like `hostnamectl` and `whoami` to ensure you are remotely accessing your spark.
- Specify a file path or directory and start editing/writing files
- Install extensions
- Clone repos
- Locally host LLM code assistant
- Verify that you can access your Spark's filesystem with VS Code as a text editor
- Open the integrated terminal in VS Code and run test commands like `hostnamectl` and `whoami` to ensure you are remotely accessing your Spark
- Navigate to a specific file path or directory and start editing/writing files
- Install VS Code extensions for your development workflow (Python, Docker, GitLens, etc.)
- Clone repositories from GitHub or other version control systems
- Configure and locally host an LLM code assistant if desired