The latest release of the open-source development system introduces a significant step forward for developers working with complex, multi-core environments and artificial intelligence.
Analog Devices has launched CodeFusion Studio 2.0, upgrading its open-source embedded development platform with comprehensive AI workflow support and unified multi-core orchestration capabilities. The Visual Studio Code-based environment now enables engineers to develop, debug, and deploy across ADI’s entire processor and microcontroller portfolio from a single workspace, addressing the fragmentation that has long plagued multi-core embedded development.
Technical enhancements
The platform introduces end-to-end AI model deployment with bring-your-own-model capability, allowing engineers to import, convert, and profile models directly on ADI hardware without switching between toolchains. A new Zephyr-based modular framework provides runtime AI/ML profiling with layer-by-layer performance analysis, operator counts, memory watermarks, and CPU utilization metrics. The framework supports multiple backends, including TensorFlow Lite Micro and TVM, with automatic code generation optimized for specific ADI silicon targets.
Multi-core development sees significant improvements with unified debugging across CPUs, DSPs, and NPUs. Engineers can now trace inter-core interactions, inspect shared resources, and analyze crashes through integrated core dump analysis—all within one environment. The System Planner provides graphical SOC visualization with automated validation of shared resources and cross-core communication paths. Core assignment for AI models happens automatically, with the platform handling memory allocation and inter-processor communication setup.
“The instrumentation hooks directly into TensorFlow Lite Micro and TVM frameworks to extract telemetry during inference,” explained David Boland, Senior Director of Software Tools and Frameworks at ADI. “We pipe that through Zephyr’s instrumentation system to provide detailed performance metrics, including activation counts and memory usage per layer.”
Developer productivity features
CodeFusion Studio 2.0 consolidates previously fragmented workflows that required multiple IDEs per core type. The unified workspace maintains consistent build dependencies, shared memory maps, and peripheral management across all cores. Advanced GDB integration supports custom JSON and Python scripting for multi-core debugging on both Windows and Linux platforms, with RTOS thread awareness providing visibility into task scheduling and synchronization issues.
The platform has expanded hardware support from 5 to 24 ADI products in one year, spanning from ultra-low-power edge devices to high-performance DSPs. An integrated package manager enables on-demand SDK and toolchain downloads with optional telemetry for diagnostics. The AutoML plugin accelerates model exploration and tuning workflows, while a built-in model compatibility checker validates deployment feasibility before runtime.
New debugging capabilities include a System Event Viewer that functions as an integrated logic analyzer, enabling engineers to monitor events, set triggers, and stream data for timing analysis across cores. An ELF file explorer provides graphical memory and flash usage mapping, helping teams optimize resource allocation. The platform also features automated board bring-up with simplified SDK installation and example code deployment.
CodeFusion Studio 2.0 is available now at developer.analog.com/solutions/codefusionstudio. The platform maintains backward compatibility with existing ADI development boards and evaluation kits while adding support for newer multi-core SOCs and edge AI accelerators. Future updates will expand runtime environments and deepen hardware-software integration as ADI continues developing its Physical Intelligence ecosystem.
Leave a Reply