#!/bin/bash # RusTorch GPU-Enabled Rust Kernel Quick Start Script - Multilingual Support # Auto-detects system language for international users # Supports: English, Japanese, Spanish, French, German, Chinese, Korean # # Usage: curl -sSL https://raw.githubusercontent.com/JunSuzukiJapan/rustorch/main/quick_start_rust_kernel_gpu.sh | bash # 使用法: curl -sSL https://raw.githubusercontent.com/JunSuzukiJapan/rustorch/main/quick_start_rust_kernel_gpu.sh | bash set -e # Language Detection and Message System detect_language() { local lang_code # Try multiple methods to detect language if [[ -n "${LC_ALL:-}" ]]; then lang_code="${LC_ALL%.*}" elif [[ -n "${LC_MESSAGES:-}" ]]; then lang_code="${LC_MESSAGES%.*}" elif [[ -n "${LANG:-}" ]]; then lang_code="${LANG%.*}" else lang_code="en_US" fi # Extract language prefix lang_code="${lang_code%_*}" case "$lang_code" in ja) echo "ja" ;; es) echo "es" ;; fr) echo "fr" ;; de) echo "de" ;; zh|zh_CN|zh_TW) echo "zh" ;; ko) echo "ko" ;; *) echo "en" ;; esac } # Multilingual message function for GPU Rust Kernel msg() { local key="$1" local lang="${DETECTED_LANG:-en}" case "$key" in "welcome_title") case "$lang" in en) echo "🦀🚀 RusTorch GPU-Enabled Rust Kernel Quick Start" ;; ja) echo "🦀🚀 RusTorch GPU対応 Rust カーネル クイックスタート" ;; es) echo "🦀🚀 Inicio Rápido Kernel Rust GPU RusTorch" ;; fr) echo "🦀🚀 Démarrage Rapide Noyau Rust GPU RusTorch" ;; de) echo "🦀🚀 RusTorch GPU Rust Kernel Schnellstart" ;; zh) echo "🦀🚀 RusTorch GPU Rust 内核快速开始" ;; ko) echo "🦀🚀 RusTorch GPU Rust 커널 빠른 시작" ;; esac ;; "gpu_detection") case "$lang" in en) echo "🎮 Detecting GPU capabilities..." ;; ja) echo "🎮 GPU性能を検出中..." ;; es) echo "🎮 Detectando capacidades de GPU..." ;; fr) echo "🎮 Détection des capacités GPU..." ;; de) echo "🎮 GPU-Fähigkeiten erkennen..." ;; zh) echo "🎮 检测 GPU 功能..." ;; ko) echo "🎮 GPU 기능 감지 중..." ;; esac ;; "gpu_kernel_complete") case "$lang" in en) echo "🎉 GPU-Enabled Rust Kernel Setup Complete!" ;; ja) echo "🎉 GPU対応Rustカーネルセットアップ完了！" ;; es) echo "🎉 ¡Configuración Kernel Rust GPU Completa!" ;; fr) echo "🎉 Configuration Noyau Rust GPU Terminée!" ;; de) echo "🎉 GPU Rust Kernel Setup Abgeschlossen!" ;; zh) echo "🎉 GPU Rust 内核设置完成！" ;; ko) echo "🎉 GPU Rust 커널 설정 완료！" ;; esac ;; "detected_gpu_features") case "$lang" in en) echo "🎮 Detected GPU features:" ;; ja) echo "🎮 検出されたGPU機能:" ;; es) echo "🎮 Características GPU detectadas:" ;; fr) echo "🎮 Fonctionnalités GPU détectées:" ;; de) echo "🎮 Erkannte GPU-Features:" ;; zh) echo "🎮 检测到的 GPU 功能：" ;; ko) echo "🎮 감지된 GPU 기능:" ;; esac ;; "gpu_demo_loaded") case "$lang" in en) echo "📝 GPU Demo notebook loaded automatically" ;; ja) echo "📝 GPUデモノートブック自動読み込み済み" ;; es) echo "📝 Notebook demo GPU cargado automáticamente" ;; fr) echo "📝 Notebook de démo GPU chargé automatiquement" ;; de) echo "📝 GPU-Demo-Notebook automatisch geladen" ;; zh) echo "📝 GPU 演示笔记本自动加载" ;; ko) echo "📝 GPU 데모 노트북 자동 로드됨" ;; esac ;; esac } # Detect system language DETECTED_LANG=$(detect_language) # Display welcome message in user's language echo "$(msg "welcome_title")" echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "" echo "🌍 Language detected: $DETECTED_LANG" echo "" # Create workspace RUSTORCH_DIR="$HOME/rustorch-gpu-rust-kernel" echo "📁 Creating RusTorch GPU Rust workspace: $RUSTORCH_DIR" if [ -d "$RUSTORCH_DIR" ]; then echo "⚠️ Directory exists. Updating..." cd "$RUSTORCH_DIR" git pull origin main || echo "Git pull failed, continuing..." else echo "📥 Downloading RusTorch..." git clone --depth 1 https://github.com/JunSuzukiJapan/rustorch.git "$RUSTORCH_DIR" cd "$RUSTORCH_DIR" fi echo "" echo "🔍 Checking system requirements..." # Check Rust if command -v rustc &> /dev/null; then RUST_VERSION=$(rustc --version | cut -d' ' -f2) echo "✅ Rust: $RUST_VERSION" else echo "📦 Installing Rust..." curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y source $HOME/.cargo/env echo "✅ Rust installed!" fi # Check Python if command -v python3 &> /dev/null; then PYTHON_VERSION=$(python3 --version | cut -d' ' -f2) echo "✅ Python: $PYTHON_VERSION" else echo "❌ Python 3 is required" exit 1 fi # GPU Detection and Feature Configuration echo "" echo "$(msg "gpu_detection")" GPU_FEATURES="" GPU_FOUND=false # NVIDIA CUDA Detection if command -v nvidia-smi &> /dev/null; then CUDA_VERSION=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader,nounits | head -1) echo "✅ NVIDIA GPU detected (Driver: $CUDA_VERSION)" GPU_FEATURES="cuda" GPU_FOUND=true elif [[ -d "/usr/local/cuda" ]] || [[ -d "/opt/cuda" ]]; then echo "✅ CUDA installation detected" GPU_FEATURES="cuda" GPU_FOUND=true fi # Apple Metal Detection (macOS) if [[ "$OSTYPE" == "darwin"* ]]; then METAL_CHECK=$(system_profiler SPDisplaysDataType 2>/dev/null | grep -i "metal" || echo "") if [[ -n "$METAL_CHECK" ]]; then echo "✅ Apple Metal GPU detected" if [[ -n "$GPU_FEATURES" ]]; then GPU_FEATURES="$GPU_FEATURES,metal" else GPU_FEATURES="metal" fi GPU_FOUND=true fi fi # OpenCL Detection if command -v clinfo &> /dev/null; then OPENCL_DEVICES=$(clinfo -l 2>/dev/null | grep -c "Platform #" || echo "0") if [[ "$OPENCL_DEVICES" -gt 0 ]]; then echo "✅ OpenCL GPU detected ($OPENCL_DEVICES platforms)" if [[ -n "$GPU_FEATURES" ]]; then GPU_FEATURES="$GPU_FEATURES,opencl" else GPU_FEATURES="opencl" fi GPU_FOUND=true fi fi if [[ "$GPU_FOUND" == false ]]; then echo "⚠️ No GPU acceleration detected. Using CPU-only mode." echo " 💡 For GPU support, install:" echo " - NVIDIA: CUDA toolkit" echo " - Apple: Already available on macOS" echo " - Generic: OpenCL drivers" GPU_FEATURES="" else echo "🚀 GPU features available: $GPU_FEATURES" fi # Check ZMQ (required for evcxr_jupyter) echo "" echo "🔧 Checking ZMQ dependency..." if command -v pkg-config &> /dev/null && pkg-config --exists libzmq; then echo "✅ ZMQ already installed" elif [[ "$OSTYPE" == "darwin"* ]]; then if command -v brew &> /dev/null; then echo "📦 Installing ZMQ via Homebrew..." brew install zmq pkg-config else echo "❌ Please install Homebrew first: https://brew.sh" exit 1 fi elif [[ "$OSTYPE" == "linux-gnu"* ]]; then echo "📦 Installing ZMQ for Linux..." if command -v apt-get &> /dev/null; then sudo apt-get update && sudo apt-get install -y libzmq3-dev pkg-config elif command -v yum &> /dev/null; then sudo yum install -y zeromq-devel pkgconfig else echo "❌ Please install libzmq3-dev manually" exit 1 fi else echo "⚠️ Unsupported OS for automatic ZMQ installation" fi echo "" echo "🦀 Installing Rust Jupyter kernel (evcxr_jupyter)..." # Install evcxr_jupyter if command -v evcxr_jupyter &> /dev/null; then echo "✅ evcxr_jupyter already installed" else echo "📦 Installing evcxr_jupyter..." cargo install evcxr_jupyter fi # Install the kernel echo "🔧 Installing Rust kernel for Jupyter..." evcxr_jupyter --install echo "" echo "📚 Installing Python dependencies..." if [ ! -d ".venv" ]; then python3 -m venv .venv fi source .venv/bin/activate pip install --upgrade pip jupyter jupyterlab matplotlib pandas numpy # Create GPU-enabled Rust kernel demo notebook echo "" echo "📝 Creating GPU-enabled Rust kernel demo notebook..." mkdir -p notebooks cat > notebooks/rustorch_gpu_rust_kernel_demo.ipynb << 'EOF' { "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# 🦀🚀 RusTorch GPU-Enabled Rust Kernel Demo\n", "# 🦀🚀 RusTorch GPU対応 Rustカーネルデモ\n", "\n", "This notebook demonstrates how to use RusTorch with GPU acceleration directly in Rust within Jupyter!\n", "\n", "このノートブックでは、Jupyter内でRustを直接使ってGPU加速されたRusTorchを使用する方法を示します！" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 📦 Install RusTorch with GPU Features\n", "## 📦 GPU機能付きRusTorchをインストール\n", "\n", "First, let's add RusTorch with GPU features as a dependency:\n", "\n", "まず、GPU機能付きRusTorchを依存関係として追加しましょう：" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "// Check which GPU features are available on this system\n", ":dep rustorch = { version = \"0.5.10\", features = [\"cuda\", \"metal\", \"opencl\"] }\n", ":dep ndarray = \"0.16\"" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 🎯 Basic Tensor Operations with GPU\n", "## 🎯 GPU基本的なテンソル操作" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "use rustorch::*;\n", "use ndarray::prelude::*;\n", "\n", "// Check GPU availability\n", "let gpu_available = rustorch::cuda::is_available() || \n", " rustorch::metal::is_available() || \n", " rustorch::opencl::is_available();\n", "\n", "println!(\"🎮 GPU acceleration available: {}\", gpu_available);\n", "\n", "// Create tensors\n", "let a = Tensor::from_array(array![[1.0, 2.0], [3.0, 4.0]]);\n", "let b = Tensor::from_array(array![[5.0, 6.0], [7.0, 8.0]]);\n", "\n", "println!(\"Tensor a: {:?}\", a);\n", "println!(\"Tensor b: {:?}\", b);" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "// Move tensors to GPU if available\n", "let device = if rustorch::cuda::is_available() {\n", " println!(\"🚀 Using CUDA GPU acceleration\");\n", " Device::Cuda(0)\n", "} else if rustorch::metal::is_available() {\n", " println!(\"🍎 Using Apple Metal GPU acceleration\");\n", " Device::Metal\n", "} else if rustorch::opencl::is_available() {\n", " println!(\"⚡ Using OpenCL GPU acceleration\");\n", " Device::OpenCL(0)\n", "} else {\n", " println!(\"💻 Using CPU (no GPU available)\");\n", " Device::Cpu\n", "};\n", "\n", "let a_gpu = a.to_device(&device);\n", "let b_gpu = b.to_device(&device);\n", "\n", "// GPU-accelerated matrix multiplication\n", "let result = a_gpu.matmul(&b_gpu);\n", "println!(\"GPU matrix multiplication result: {:?}\", result);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 🧮 GPU-Accelerated Advanced Operations\n", "## 🧮 GPU加速高度な操作" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "// Create larger tensors for GPU performance benefits\n", "let large_tensor = Tensor::randn(&[1000, 1000]).to_device(&device);\n", "println!(\"Large tensor shape: {:?}\", large_tensor.shape());\n", "println!(\"Device: {:?}\", large_tensor.device());\n", "\n", "// GPU-accelerated activation functions\n", "let relu_result = large_tensor.relu();\n", "println!(\"GPU ReLU completed on {}x{} tensor\", 1000, 1000);\n", "\n", "let sigmoid_result = large_tensor.sigmoid();\n", "println!(\"GPU Sigmoid completed on {}x{} tensor\", 1000, 1000);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 🤖 GPU-Accelerated Neural Network\n", "## 🤖 GPU加速ニューラルネットワーク" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "use rustorch::nn::*;\n", "\n", "// Create a neural network on GPU\n", "let mut network = Sequential::new()\n", " .add_layer(Linear::new(784, 256).to_device(&device))\n", " .add_layer(ReLU::new())\n", " .add_layer(Linear::new(256, 128).to_device(&device))\n", " .add_layer(ReLU::new())\n", " .add_layer(Linear::new(128, 10).to_device(&device))\n", " .add_layer(Softmax::new());\n", "\n", "println!(\"🧠 GPU Neural network created with {} parameters\", network.num_parameters());\n", "println!(\"🎮 Network device: {:?}\", device);\n", "\n", "// Create sample input on GPU\n", "let input = Tensor::randn(&[32, 784]).to_device(&device); // Batch size 32\n", "let output = network.forward(&input);\n", "\n", "println!(\"Input shape: {:?}\", input.shape());\n", "println!(\"Output shape: {:?}\", output.shape());\n", "println!(\"✅ GPU forward pass completed!\");" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## ⚡ GPU Performance Benchmarks\n", "## ⚡ GPUパフォーマンスベンチマーク" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "use std::time::Instant;\n", "\n", "// Benchmark CPU vs GPU performance\n", "let size = 1000;\n", "let iterations = 5;\n", "\n", "println!(\"🏁 Benchmarking {}x{} matrix operations ({} iterations)...\", size, size, iterations);\n", "println!(\"\");\n", "\n", "// CPU benchmark\n", "println!(\"💻 CPU Benchmark:\");\n", "let a_cpu = Tensor::randn(&[size, size]);\n", "let b_cpu = Tensor::randn(&[size, size]);\n", "\n", "let start = Instant::now();\n", "for i in 0..iterations {\n", " let _result = a_cpu.matmul(&b_cpu);\n", " if i == 0 { println!(\" Iteration {} completed\", i + 1); }\n", "}\n", "let cpu_duration = start.elapsed();\n", "let cpu_gflops = (2.0 * size as f64 * size as f64 * size as f64 * iterations as f64) \n", " / (cpu_duration.as_secs_f64() * 1e9);\n", "\n", "println!(\" ⏱️ CPU Time: {:?}\", cpu_duration);\n", "println!(\" 📊 CPU GFLOPS: {:.2}\", cpu_gflops);" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "// GPU benchmark (if available)\n", "if gpu_available {\n", " println!(\"\");\n", " println!(\"🎮 GPU Benchmark:\");\n", " let a_gpu = Tensor::randn(&[size, size]).to_device(&device);\n", " let b_gpu = Tensor::randn(&[size, size]).to_device(&device);\n", " \n", " // Warm up GPU\n", " let _ = a_gpu.matmul(&b_gpu);\n", " \n", " let start = Instant::now();\n", " for i in 0..iterations {\n", " let _result = a_gpu.matmul(&b_gpu);\n", " if i == 0 { println!(\" Iteration {} completed\", i + 1); }\n", " }\n", " let gpu_duration = start.elapsed();\n", " let gpu_gflops = (2.0 * size as f64 * size as f64 * size as f64 * iterations as f64) \n", " / (gpu_duration.as_secs_f64() * 1e9);\n", " \n", " println!(\" ⏱️ GPU Time: {:?}\", gpu_duration);\n", " println!(\" 📊 GPU GFLOPS: {:.2}\", gpu_gflops);\n", " \n", " if gpu_duration < cpu_duration {\n", " let speedup = cpu_duration.as_secs_f64() / gpu_duration.as_secs_f64();\n", " println!(\" 🚀 GPU Speedup: {:.2}x faster!\", speedup);\n", " }\n", "} else {\n", " println!(\"\");\n", " println!(\"⚠️ GPU benchmark skipped (no GPU acceleration available)\");\n", "}" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 🔥 Advanced GPU Examples\n", "## 🔥 高度なGPUの例" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "if gpu_available {\n", " // Convolutional layer example on GPU\n", " println!(\"🧠 Creating GPU-accelerated CNN layer...\");\n", " \n", " // Create sample image tensor (batch_size=4, channels=3, height=32, width=32)\n", " let image_batch = Tensor::randn(&[4, 3, 32, 32]).to_device(&device);\n", " \n", " // Create Conv2D layer on GPU\n", " let conv_layer = Conv2d::new(3, 16, 3) // 3 input channels, 16 output, 3x3 kernel\n", " .padding(1)\n", " .to_device(&device);\n", " \n", " let conv_start = Instant::now();\n", " let conv_output = conv_layer.forward(&image_batch);\n", " let conv_duration = conv_start.elapsed();\n", " \n", " println!(\"📐 Input shape: {:?}\", image_batch.shape());\n", " println!(\"📐 Output shape: {:?}\", conv_output.shape());\n", " println!(\"⏱️ GPU Convolution time: {:?}\", conv_duration);\n", " \n", " // Batch processing example\n", " println!(\"\");\n", " println!(\"📊 Batch processing with GPU acceleration...\");\n", " \n", " let batch_size = 128;\n", " let feature_size = 512;\n", " let large_batch = Tensor::randn(&[batch_size, feature_size]).to_device(&device);\n", " \n", " let batch_start = Instant::now();\n", " let normalized = large_batch.layer_norm(&[feature_size]);\n", " let activated = normalized.gelu();\n", " let batch_duration = batch_start.elapsed();\n", " \n", " println!(\"✅ Processed batch of {} samples in {:?}\", batch_size, batch_duration);\n", " println!(\"📈 Throughput: {:.0} samples/second\", \n", " batch_size as f64 / batch_duration.as_secs_f64());\n", "} else {\n", " println!(\"⚠️ Advanced GPU examples skipped (GPU acceleration not available)\");\n", " println!(\"💡 To enable GPU acceleration:\");\n", " println!(\" - Install CUDA drivers for NVIDIA GPUs\");\n", " println!(\" - Use macOS for Apple Metal support\");\n", " println!(\" - Install OpenCL drivers for other GPUs\");\n", "}" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 🎉 Conclusion\n", "## 🎉 まとめ\n", "\n", "You can now write and execute GPU-accelerated Rust code directly in Jupyter!\n", "\n", "これでJupyter内で直接GPU加速されたRustコードを書いて実行できます！\n", "\n", "**Benefits / 利点:**\n", "- 🚀 Native Rust performance / ネイティブRustパフォーマンス\n", "- 🎮 GPU acceleration / GPU加速\n", "- 🔧 Direct library access / ライブラリへの直接アクセス\n", "- 🎯 Type safety / 型安全性\n", "- ⚡ Zero-cost abstractions / ゼロコスト抽象化\n", "- 🧠 Advanced ML operations / 高度なML操作" ] } ], "metadata": { "kernelspec": { "display_name": "Rust", "language": "rust", "name": "rust" }, "language_info": { "codemirror_mode": "rust", "file_extension": ".rs", "mimetype": "text/rust", "name": "Rust", "pygment_lexer": "rust", "version": "" } }, "nbformat": 4, "nbformat_minor": 4 } EOF echo "" echo "🎯 Starting Jupyter Lab with GPU-enabled Rust kernel..." jupyter lab --port=8889 --no-browser notebooks/rustorch_gpu_rust_kernel_demo.ipynb & echo "" echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "$(msg "gpu_kernel_complete")" echo "" echo "$(msg "how_to_use")" echo " $(msg "select_rust_kernel")" echo " $(msg "write_rust_code")" echo " 3. Use GPU features with detected capabilities" echo "" echo "$(msg "detected_gpu_features") $GPU_FEATURES" echo "" echo "$(msg "available_at")" echo "" echo "$(msg "gpu_demo_loaded")" echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"