From b046b203f610d6d52c2006857894d58ebd5081c9 Mon Sep 17 00:00:00 2001 From: Siddartha Pothapragada Date: Wed, 15 Oct 2025 08:54:17 -0700 Subject: [PATCH] Summary: Add the Cross compilation Script for RPi (4 & 5) for Linux host machine (#15014) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Test Plan: examples/raspberry_pi/rpi_setup.sh pi5 ... [100%] Linking CXX executable llama_main [100%] Built target llama_main [SUCCESS] LLaMA runner built successfully ==== Extracting Bundled Libraries ==== [INFO] Extracting GLIBC libraries from toolchain... [WARNING] Use bundled GLIBC script on RPI device ONLY if you encounter a GLIBC mismatch error when running llama_main. [SUCCESS] Bundled libraries prepared in: /home/sidart/working/executorch/cmake-out/bundled-libs [INFO] On Raspberry Pi, run: sudo ./install_libs.sh ==== Verifying Build Outputs ==== [INFO] Checking required binaries... [SUCCESS] āœ“ llama_main (6.1M) [SUCCESS] āœ“ libllama_runner.so (4.0M) [SUCCESS] āœ“ libextension_module.a (89K) - static library [SUCCESS] All required binaries built successfully! ==== Setup Complete! ==== āœ“ ExecuTorch cross-compilation setup completed successfully! šŸ“¦ Built binaries: ā€¢ llama_main: /home/sidart/working/executorch/cmake-out/examples/models/llama/llama_main ā€¢ libllama_runner.so: /home/sidart/working/executorch/cmake-out/examples/models/llama/runner/libllama_runner.so ā€¢ libextension_module.a: Statically linked into llama_main āœ… ā€¢ Bundled libraries: /home/sidart/working/executorch/cmake-out/bundled-libs/ šŸ“‹ Next steps: 1. Copy binaries to your Raspberry Pi pi5: scp /home/sidart/working/executorch/cmake-out/examples/models/llama/llama_main pi@:~/ scp /home/sidart/working/executorch/cmake-out/examples/models/llama/runner/libllama_runner.so pi@:~/ scp -r /home/sidart/working/executorch/cmake-out/bundled-libs/ pi@:~/ 2. Copy shared libraries to system location: sudo cp libllama_runner.so /lib/ # Only this one needed! sudo ldconfig 3. Dry run to check for GLIBC or other issues: ./llama_main --help # Ensure there are no GLIBC or other errors before proceeding. 4. If you see GLIBC errors, install bundled libraries: cd ~/bundled-libs && sudo ./install_libs.sh source setup_env.sh # Only do this if you encounter a GLIBC version mismatch or similar error. 5. Download your model and tokenizer: # Refer to the official documentation for exact details. 6. Run ExecuTorch with your model: ./llama_main --model_path ./model.pte --tokenizer_path ./tokenizer.model --seq_len 128 --prompt "What is the meaning of life ?" šŸŽÆ Deployment Summary: šŸ“ Files to copy: 2 (llama_main + libllama_runner.so) šŸ—ļø Extension module: Built-in (no separate .so needed) šŸ”§ Toolchain saved at: /home/sidart/working/executorch/arm-toolchain/arm-gnu-toolchain-14.3.rel1-x86_64-aarch64-none-linux-gnu šŸ”§ CMake toolchain file: /home/sidart/working/executorch/arm-toolchain-pi5.cmake Happy inferencing! šŸš€ ### Summary [PLEASE REMOVE] See [CONTRIBUTING.md's Pull Requests](https://github.com/pytorch/executorch/blob/main/CONTRIBUTING.md#pull-requests) for ExecuTorch PR guidelines. [PLEASE REMOVE] If this PR closes an issue, please add a `Fixes #` line. [PLEASE REMOVE] If this PR introduces a fix or feature that should be the upcoming release notes, please add a "Release notes: " label. For a list of available release notes labels, check out [CONTRIBUTING.md's Pull Requests](https://github.com/pytorch/executorch/blob/main/CONTRIBUTING.md#pull-requests). ### Test plan [PLEASE REMOVE] How did you test this PR? Please write down any manual commands you used and note down tests that you have written if applicable. (cherry picked from commit a0e5280368e14cc2fa0cc4f42cf68224f4e88286) --- .../pico2/CMakeLists.txt | 0 .../{rpi => raspberry_pi}/pico2/README.md | 8 +- .../pico2}/build_firmware_pico.sh | 0 examples/{rpi => raspberry_pi}/pico2/main.cpp | 0 .../{rpi => raspberry_pi}/pico2/model_pte.c | 0 .../{rpi => raspberry_pi}/pico2/model_pte.h | 0 .../pico2}/pte_to_array.py | 0 examples/raspberry_pi/setup.sh | 616 ++++++++++++++++++ 8 files changed, 623 insertions(+), 1 deletion(-) rename examples/{rpi => raspberry_pi}/pico2/CMakeLists.txt (100%) rename examples/{rpi => raspberry_pi}/pico2/README.md (89%) rename examples/{rpi => raspberry_pi/pico2}/build_firmware_pico.sh (100%) rename examples/{rpi => raspberry_pi}/pico2/main.cpp (100%) rename examples/{rpi => raspberry_pi}/pico2/model_pte.c (100%) rename examples/{rpi => raspberry_pi}/pico2/model_pte.h (100%) rename examples/{rpi => raspberry_pi/pico2}/pte_to_array.py (100%) create mode 100755 examples/raspberry_pi/setup.sh diff --git a/examples/rpi/pico2/CMakeLists.txt b/examples/raspberry_pi/pico2/CMakeLists.txt similarity index 100% rename from examples/rpi/pico2/CMakeLists.txt rename to examples/raspberry_pi/pico2/CMakeLists.txt diff --git a/examples/rpi/pico2/README.md b/examples/raspberry_pi/pico2/README.md similarity index 89% rename from examples/rpi/pico2/README.md rename to examples/raspberry_pi/pico2/README.md index 8c617cc5aa3..44069cb659e 100644 --- a/examples/rpi/pico2/README.md +++ b/examples/raspberry_pi/pico2/README.md @@ -2,8 +2,10 @@ This document outlines the steps required to run a simple Add Module on the Pico2 microcontroller using executorch. ## (Pre-requisistes) Prepare the Environment for Arm -1. See for instructions on setting up the environment for Arm. + +1. Setup executorch development environment, Also see for instructions on setting up the environment for Arm. 2. Make sure you have the toolchain configured correctly. + ```bash which arm-none-eabi-gcc --> return something like executorch/examples/arm/ethos-u-scratch/arm-gnu-toolchain-13.3.rel1-x86_64-arm-none-eabi/bin/arm-none-eabi-gcc @@ -22,9 +24,11 @@ executorch/examples/rpi/build_firmware_pico.sh --model= ``` ### Flash Firmware + Hold the BOOTSEL button on the Pico2 and connect it to your computer; it will mount as RPI-RP2. Copy the executorch_pico.uf2 file to this drive. ### Verify Execution + Check that the Pico2's LED blinks 10 times at 500 ms interval to confirm successful firmware execution. The Pico2's LED should blink 10 times at 500 ms intervals, indicating successful firmware execution. If connected via serial, you should see: @@ -34,7 +38,9 @@ Output: 13.000000, 136.000000, 24.000000, 131.000000 ``` ### Debugging via Serial Terminal + On macOS or Linux, open a serial terminal with: + ```bash screen /dev/tty.usbmodem1101 115200 ``` diff --git a/examples/rpi/build_firmware_pico.sh b/examples/raspberry_pi/pico2/build_firmware_pico.sh similarity index 100% rename from examples/rpi/build_firmware_pico.sh rename to examples/raspberry_pi/pico2/build_firmware_pico.sh diff --git a/examples/rpi/pico2/main.cpp b/examples/raspberry_pi/pico2/main.cpp similarity index 100% rename from examples/rpi/pico2/main.cpp rename to examples/raspberry_pi/pico2/main.cpp diff --git a/examples/rpi/pico2/model_pte.c b/examples/raspberry_pi/pico2/model_pte.c similarity index 100% rename from examples/rpi/pico2/model_pte.c rename to examples/raspberry_pi/pico2/model_pte.c diff --git a/examples/rpi/pico2/model_pte.h b/examples/raspberry_pi/pico2/model_pte.h similarity index 100% rename from examples/rpi/pico2/model_pte.h rename to examples/raspberry_pi/pico2/model_pte.h diff --git a/examples/rpi/pte_to_array.py b/examples/raspberry_pi/pico2/pte_to_array.py similarity index 100% rename from examples/rpi/pte_to_array.py rename to examples/raspberry_pi/pico2/pte_to_array.py diff --git a/examples/raspberry_pi/setup.sh b/examples/raspberry_pi/setup.sh new file mode 100755 index 00000000000..0b4f43ed1d3 --- /dev/null +++ b/examples/raspberry_pi/setup.sh @@ -0,0 +1,616 @@ +#!/usr/bin/env bash +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under the BSD-style license found in the +# LICENSE file in the root directory of this source tree. + + +# Prerequisites: Create a virtual environment using conda or venv and activate it. +# setup.sh - ExecuTorch Raspberry Pi Cross-Compilation Setup +# Usage: ./setup.sh [pi4|pi5] + + +set -euo pipefail + +# Configuration +SCRIPT_VERSION="1.0" +TOOLCHAIN_VERSION="14.3.rel1" +WORKSPACE_DIR="$(pwd)" +TOOLCHAIN_DIR="${WORKSPACE_DIR}/examples/raspberry_pi/gen/arm-toolchain" +CMAKE_OUT_DIR="${WORKSPACE_DIR}/cmake-out" + +FORCE_REBUILD="${FORCE_REBUILD:-false}" +FORCE_DOWNLOAD="${FORCE_DOWNLOAD:-false}" +SKIP_ENV_SETUP="${SKIP_ENV_SETUP:-false}" + +# Colors for output +RED='\033[0;31m' +GREEN='\033[0;32m' +YELLOW='\033[1;33m' +BLUE='\033[0;34m' +NC='\033[0m' # No Color + +# Default values +RPI_MODEL="" +HOST_OS="" +HOST_ARCH="" + +# Logging functions +log_info() { + echo -e "${BLUE}[INFO]${NC} $1" +} + +log_success() { + echo -e "${GREEN}[SUCCESS]${NC} $1" +} + +log_warning() { + echo -e "${YELLOW}[WARNING]${NC} $1" +} + +log_error() { + echo -e "${RED}[ERROR]${NC} $1" +} + +log_step() { + echo -e "\n${GREEN}==== $1 ====${NC}" +} + +# Parse command line arguments +parse_args() { + while [[ $# -gt 0 ]]; do + case $1 in + pi4|pi5) + RPI_MODEL="$1" + shift + ;; + --force-rebuild) + FORCE_REBUILD=true + shift + ;; + --force-download) + FORCE_DOWNLOAD=true + shift + ;; + --skip-env-setup) + SKIP_ENV_SETUP=true + shift + ;; + --clean) + CLEAN=true + shift + ;; + -h|--help) + show_help + exit 0 + ;; + *) + log_error "Unknown option: $1" + show_help + exit 1 + ;; + esac + done + + if [[ -z "$RPI_MODEL" ]]; then + log_error "Please specify Raspberry Pi model: pi4 or pi5" + show_help + exit 1 + fi +} + +show_help() { + cat << EOF +ExecuTorch Raspberry Pi Cross-Compilation Setup v${SCRIPT_VERSION} + +Usage: $0 [pi4|pi5] [options] + +Arguments: + pi4|pi5 Target Raspberry Pi model + +Options: + --force-rebuild Force complete rebuild of all components + --force-download Force re-download of the ARM toolchain (default: disabled) + --skip-env-setup Skip Python environment setup (just rebuild code) + --clean Remove build artifacts and exit + -h, --help Show this help message + +Examples: + $0 pi5 # Setup for Pi5 + $0 pi4 # Setup for Pi4 + $0 pi5 --force-rebuild # Force complete rebuild + $0 pi5 --skip-env-setup --force-rebuild # Just rebuild code, skip env setup + $0 pi5 --force-download # Force re-download toolchain + $0 pi5 --clean # Clean up build artifacts and exit + +EOF +} + +# Clean up build artifacts +clean_up() { + log_step "Cleaning up build artifacts" + rm -rf "$TOOLCHAIN_DIR" "$CMAKE_OUT_DIR" arm-toolchain-*.cmake + log_success "Cleanup complete. Exiting." + exit 0 +} + +# Detect host system +detect_host() { + log_step "Detecting Host System" + + HOST_OS=$(uname -s) + HOST_ARCH=$(uname -m) + + case "$HOST_OS" in + Linux) + log_info "Detected Linux host" + ;; + Darwin) + HOST_OS="Darwin" + log_info "Detected macOS host" + log_warning "Cross compilation on mac host is WIP" + exit 1 + ;; + *) + log_error "Unsupported host OS: $HOST_OS" + log_error "This script supports Linux and macOS only" + exit 1 + ;; + esac + + case "$HOST_ARCH" in + x86_64|amd64|arm64|aarch64) + HOST_ARCH="x86_64" + ;; + *) + log_error "Unsupported host architecture: $HOST_ARCH" + exit 1 + ;; + esac + + log_success "Host: $HOST_OS $HOST_ARCH" +} + +# Check prerequisites +check_prerequisites() { + log_step "Checking Prerequisites" + + # Check required tools + local required_tools=("cmake" "git" "curl" "tar" "md5sum") + + for tool in "${required_tools[@]}"; do + if ! command -v "$tool" &> /dev/null; then + log_error "$tool is not installed" + case "$HOST_OS" in + Linux) + log_info "Install with: sudo apt-get install $tool (Ubuntu/Debian) or equivalent" + ;; + Darwin) + log_info "Install with: brew install $tool" + ;; + esac + exit 1 + fi + done + + # Check CMake version + local cmake_version=$(cmake --version | head -n1 | cut -d' ' -f3) + local required_cmake="3.19" + + if ! command -v python3 -c "import sys; sys.exit(0 if tuple(map(int, '$cmake_version'.split('.'))) >= tuple(map(int, '$required_cmake'.split('.'))) else 1)" 2>/dev/null; then + log_warning "CMake version $cmake_version detected. Recommended: $required_cmake+" + fi + + # Check if we're in ExecuTorch directory + if [[ ! -f "CMakeLists.txt" ]] || [[ ! -d "examples/models/llama" ]]; then + log_error "Not in ExecuTorch root directory" + log_info "Please run this script from the ExecuTorch root directory" + exit 1 + fi + + log_success "All prerequisites satisfied" +} + +# Detect and setup Python environment +setup_environment() { + log_step "Setting Up Python Environment" + + if [[ "$SKIP_ENV_SETUP" == "true" ]]; then + log_info "Skipping Python environment setup" + return 0 + fi + + # Check if ExecuTorch is already installed + if python3 -c "import executorch" 2>/dev/null; then + log_success "ExecuTorch already installed and importable" + return 0 + else + log_warning "ExecuTorch not found in Python environment" + fi + + # Check if already in virtual environment + if [[ -n "${VIRTUAL_ENV:-}" ]] || [[ -n "${CONDA_DEFAULT_ENV:-}" ]]; then + if [[ -n "${CONDA_DEFAULT_ENV:-}" ]]; then + log_success "Using existing conda environment: ${CONDA_DEFAULT_ENV}" + else + log_success "Using existing virtual environment: $(basename "${VIRTUAL_ENV:-}")" + fi + else + log_warning "No virtual environment detected" + + # Check if conda is available + if command -v conda &> /dev/null; then + log_info "Creating conda environment..." + conda create -yn executorch python=3.10.0 + eval "$(conda shell.bash hook)" + conda activate executorch + log_success "Created and activated conda environment: executorch" + elif command -v python3 &> /dev/null; then + log_info "Creating Python virtual environment..." + python3 -m venv .venv + source .venv/bin/activate + pip install --upgrade pip + log_success "Created and activated virtual environment: .venv" + else + log_error "Neither conda nor python3 found" + exit 1 + fi + fi + + # Initialize git submodules + if [[ -f "extension/llm/tokenizers/CMakeLists.txt" ]]; then + log_success "Git submodules already initialized" + else + log_info "Initializing git submodules..." + git submodule update --init --recursive + fi + + # Install ExecuTorch + log_info "Installing ExecuTorch dependencies..." + if [[ -f "./install_executorch.sh" ]]; then + ./install_executorch.sh + else + log_error "./install_executorch.sh not found" + log_info "Make sure you're in the ExecuTorch root directory" + exit 1 + fi + + log_success "Python environment setup complete" +} + +# Download and setup ARM toolchain +download_toolchain() { + log_step "Setting Up ARM Toolchain" + + local toolchain_name="arm-gnu-toolchain-${TOOLCHAIN_VERSION}-${HOST_ARCH}-aarch64-none-linux-gnu" + TOOLCHAIN_PATH="$TOOLCHAIN_DIR/$toolchain_name" + + # Check if toolchain already exists and force download not set + if [[ -f "$TOOLCHAIN_PATH/bin/aarch64-none-linux-gnu-gcc" ]] && [[ "$FORCE_DOWNLOAD" != "true" ]]; then + log_success "Toolchain already exists: $TOOLCHAIN_PATH" + return 0 + fi + + local toolchain_archive="${toolchain_name}.tar.xz" + local toolchain_url="https://developer.arm.com/-/media/Files/downloads/gnu/${TOOLCHAIN_VERSION}/binrel/${toolchain_archive}" + + # Create toolchain directory + mkdir -p "$TOOLCHAIN_DIR" + + log_info "Downloading ARM toolchain: $toolchain_name" + log_info "URL: $toolchain_url" + + cd "$TOOLCHAIN_DIR" + + # Download toolchain + if ! curl -L -o "$toolchain_archive" "$toolchain_url"; then + log_error "Failed to download toolchain" + log_info "Please check the URL or download manually from:" + log_info "https://developer.arm.com/downloads/-/arm-gnu-toolchain-downloads" + exit 1 + fi + + # Verify MD5 checksum + log_info "Verifying toolchain integrity..." + + # Define expected MD5 checksums + local expected_md5="" + case "${HOST_ARCH}-${TOOLCHAIN_VERSION}" in + x86_64-14.3.rel1) + expected_md5="b82906df613ca762058e3ca39ac2e23a" + ;; + aarch64-14.3.rel1) + expected_md5="b82906df613ca762058e3ca39ac2e23a" + ;; + *) + log_warning "No MD5 checksum available for ${HOST_ARCH}-${TOOLCHAIN_VERSION}, skipping verification" + ;; + esac + + if [[ -n "$expected_md5" ]]; then + local actual_md5=$(md5sum "$toolchain_archive" | cut -d' ' -f1) + if [[ "$actual_md5" != "$expected_md5" ]]; then + log_error "MD5 checksum mismatch!" + log_error "Expected: $expected_md5" + log_error "Got: $actual_md5" + log_error "Toolchain download may be corrupted" + exit 1 + fi + log_success "MD5 checksum verified" + fi + + # Extract toolchain + log_info "Extracting toolchain..." + tar -xf "$toolchain_archive" + + # Cleanup + rm "$toolchain_archive" + + cd "$WORKSPACE_DIR" + + # Set toolchain path + TOOLCHAIN_PATH="$TOOLCHAIN_DIR/$toolchain_name" + + # Verify critical toolchain files + local critical_files=( + "$TOOLCHAIN_PATH/bin/aarch64-none-linux-gnu-gcc" + "$TOOLCHAIN_PATH/bin/aarch64-none-linux-gnu-g++" + ) + + for file in "${critical_files[@]}"; do + if [[ ! -f "$file" ]]; then + log_error "Critical toolchain file missing: $file" + exit 1 + fi + done + + log_success "ARM toolchain ready: $TOOLCHAIN_PATH" +} + +# Create CMake toolchain file +create_toolchain_cmake() { + log_step "Creating CMake Toolchain File" + + local cmake_file="$WORKSPACE_DIR/arm-toolchain-${RPI_MODEL}.cmake" + + # Set architecture flags based on RPi model + local arch_flags + case "$RPI_MODEL" in + pi4) + arch_flags="-march=armv8-a -mtune=cortex-a72" + ;; + pi5) + arch_flags="-march=armv8.2-a+dotprod+fp16 -mtune=cortex-a76" + ;; + esac + + log_info "Creating toolchain file: $cmake_file" + log_info "Target: Raspberry Pi $RPI_MODEL" + log_info "Architecture flags: $arch_flags" + + cat > "$cmake_file" << EOF +# ARM Toolchain for Raspberry Pi $RPI_MODEL +# Generated by setup.sh v${SCRIPT_VERSION} + +set(CMAKE_SYSTEM_NAME Linux) +set(CMAKE_SYSTEM_PROCESSOR aarch64) +set(CMAKE_CROSSCOMPILING TRUE) + +# Path to ARM toolchain +set(TOOLCHAIN_PATH "$TOOLCHAIN_PATH") + +# Compilers +set(CMAKE_C_COMPILER "\${TOOLCHAIN_PATH}/bin/aarch64-none-linux-gnu-gcc") +set(CMAKE_CXX_COMPILER "\${TOOLCHAIN_PATH}/bin/aarch64-none-linux-gnu-g++") + +# Sysroot configuration +set(CMAKE_SYSROOT "\${TOOLCHAIN_PATH}/aarch64-none-linux-gnu/libc") +set(CMAKE_FIND_ROOT_PATH "\${CMAKE_SYSROOT}") + +# Architecture flags for Raspberry Pi $RPI_MODEL +set(ARCH_FLAGS "$arch_flags") + +set(CMAKE_C_FLAGS "\${CMAKE_C_FLAGS} \${ARCH_FLAGS}" CACHE STRING "" FORCE) +set(CMAKE_CXX_FLAGS "\${CMAKE_CXX_FLAGS} \${ARCH_FLAGS}" CACHE STRING "" FORCE) + +# Search paths +set(CMAKE_TRY_COMPILE_TARGET_TYPE STATIC_LIBRARY) +set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER) +set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY) +set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY) +set(CMAKE_FIND_ROOT_PATH_MODE_PACKAGE ONLY) +EOF + + CMAKE_TOOLCHAIN_FILE="$cmake_file" + log_success "Toolchain file created: $cmake_file" +} + +# Build ExecuTorch core libraries +build_executorch_core() { + log_step "Start building ExecuTorch core" + + # Check if already built (static only) + local extension_static="$CMAKE_OUT_DIR/extension/module/libextension_module.a" + + if [[ -f "$extension_static" ]]; then + log_success "Core libraries already built: $(basename "$extension_static")" + if [[ "$FORCE_REBUILD" != "true" ]]; then + log_info "Use --force-rebuild to rebuild core libraries" + return 0 + else + log_info "Force rebuilding core libraries..." + fi + fi + + # Only clean if forced or first build + if [[ "$FORCE_REBUILD" == "true" ]] || [[ ! -d "$CMAKE_OUT_DIR" ]]; then + log_info "Cleaning previous build..." + rm -rf "$CMAKE_OUT_DIR" + fi + + log_info "Building ExecuTorch core libraries..." + + # Configure + cmake --preset llm \ + -DCMAKE_TOOLCHAIN_FILE="$CMAKE_TOOLCHAIN_FILE" \ + -DCMAKE_BUILD_TYPE=Release \ + -DCMAKE_INSTALL_PREFIX="$CMAKE_OUT_DIR" \ + -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON + + # Build with proper CPU count detection for Mac/Linux + local cpu_count + if command -v nproc &> /dev/null; then + cpu_count=$(nproc) # Linux + elif command -v sysctl &> /dev/null; then + cpu_count=$(sysctl -n hw.ncpu) # Mac + else + cpu_count=4 # Fallback + fi + + cmake --build "$CMAKE_OUT_DIR" -j"$cpu_count" --target install --config Release + + log_success "Core libraries built successfully" +} + +# Build LLaMA runner +build_llama_runner() { + log_step "Building LLaMA Runner" + + log_info "Building llama_main and other libs..." + + # Configure LLaMA build + cmake -DCMAKE_TOOLCHAIN_FILE="$CMAKE_TOOLCHAIN_FILE" \ + -DCMAKE_INSTALL_PREFIX="$CMAKE_OUT_DIR" \ + -DBUILD_TESTING=OFF \ + -DCMAKE_BUILD_TYPE=Release \ + -B"$CMAKE_OUT_DIR/examples/models/llama" \ + examples/models/llama + + # Build LLaMA runner + cmake --build "$CMAKE_OUT_DIR/examples/models/llama" -j$(nproc 2>/dev/null || sysctl -n hw.ncpu 2>/dev/null || echo 4) --config Release + + log_success "LLaMA runner built successfully" +} + +# Verify build outputs +verify_build() { + log_step "Verifying Build Outputs" + + local all_good=true + + # Required files + local required_files=( + "$CMAKE_OUT_DIR/examples/models/llama/llama_main" + "$CMAKE_OUT_DIR/examples/models/llama/runner/libllama_runner.so" + ) + + log_info "Checking required binaries..." + + for file in "${required_files[@]}"; do + if [[ -f "$file" ]]; then + local size=$(ls -lh "$file" | awk '{print $5}') + log_success "āœ“ $(basename "$file") ($size)" + else + log_error "āœ— Missing: $file" + all_good=false + fi + done + + # Check extension module (static only) + local extension_static="$CMAKE_OUT_DIR/extension/module/libextension_module.a" + + if [[ -f "$extension_static" ]]; then + local size=$(ls -lh "$extension_static" | awk '{print $5}') + log_success "āœ“ libextension_module.a ($size) - statically linked" + else + log_error "āœ— Missing: extension module" + all_good=false + fi + + if [[ "$all_good" == "true" ]]; then + log_success "All required binaries built successfully!" + else + log_error "Some binaries are missing. Build may have failed." + exit 1 + fi +} + +# Display next steps +display_next_steps() { + log_step "Setup Complete!" + + echo -e "\n${GREEN}āœ“ ExecuTorch cross-compilation setup completed successfully!${NC}\n" + + echo "šŸ“¦ Built binaries:" + echo " ā€¢ llama_main: $CMAKE_OUT_DIR/examples/models/llama/llama_main" + echo " ā€¢ libllama_runner.so: $CMAKE_OUT_DIR/examples/models/llama/runner/libllama_runner.so" + echo " ā€¢ libextension_module.a: Statically linked into llama_main āœ…" + + echo -e "\nšŸ“‹ Next steps:" + + echo "1. Create deployment directory and copy binaries to your Raspberry Pi $RPI_MODEL:" + echo " ssh pi@ 'mkdir -p ~/executorch-deployment'" + echo " scp $CMAKE_OUT_DIR/examples/models/llama/llama_main pi@:~/executorch-deployment/" + echo " scp $CMAKE_OUT_DIR/examples/models/llama/runner/libllama_runner.so pi@:~/executorch-deployment/" + + echo -e "\n2. Set up library environment on Raspberry Pi:" + echo " ssh pi@" + echo " cd ~/executorch-deployment" + echo " echo 'export LD_LIBRARY_PATH=\$(pwd):\$LD_LIBRARY_PATH' > setup_env.sh" + echo " chmod +x setup_env.sh" + + echo -e "\n3. Test the deployment:" + echo " source setup_env.sh" + echo " ./llama_main --help" + echo " # Ensure there are no errors before proceeding. If there are errors, check the RaspberryPi Tutorials -> Troubleshooting section: hhttps://docs.pytorch.org/executorch/1.0/embedded-section.html#tutorials" + + echo -e "\n4. Download your model and tokenizer:" + echo " # Make sure you have downloaded the model and tokenizer files from Hugging Face or your source" + echo " # Refer to the official documentation for exact details" + + echo -e "\n5. Run ExecuTorch with your model:" + echo " ./llama_main --model_path ./model.pte --tokenizer_path ./tokenizer.model --seq_len 128 --prompt \"What is the meaning of life?\"" + + echo -e "\n${GREEN}šŸŽÆ Deployment Summary:${NC}" + echo " šŸ“ Files to deploy: 2 (llama_main + libllama_runner.so)" + echo " šŸ—ļø Extension module: Statically linked (built-in)" + echo " šŸ“‚ Target directory: ~/executorch-deployment/" + echo " šŸ”§ Library path: Local directory with LD_LIBRARY_PATH" + + echo -e "\nšŸ”§ Toolchain saved at: $TOOLCHAIN_PATH" + echo "šŸ”§ CMake toolchain file: $CMAKE_TOOLCHAIN_FILE" + echo -e "\n${GREEN}Happy inferencing! šŸš€${NC}" +} + +# Main execution +main() { + echo -e "${BLUE}" + cat << "EOF" +ā•”ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•— +ā•‘ ExecuTorch Raspberry Pi Setup ā•‘ +ā•‘ Cross-Compilation Script ā•‘ +ā•šā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā•ā• +EOF + echo -e "${NC}" + + parse_args "$@" + + # Handle clean option + if [[ "${CLEAN:-false}" == "true" ]]; then + clean_up + fi + + detect_host + check_prerequisites + setup_environment + download_toolchain + create_toolchain_cmake + build_executorch_core + build_llama_runner + verify_build + display_next_steps +} + +# Run main function +main "$@"