eLab releases open-source tools, datasets, and research artifacts to support reproducible research and education in energy-efficient computing, embedded systems, heterogeneous architectures, chiplet-based design, thermal modeling, and human-centered sensing. The following public releases provide software, datasets, and benchmarks developed through our research projects.
MFIT: Multi-Fidelity Thermal Modeling for 2.5D and 3D Multi-Chiplet Architectures
MFIT provides thermal RC and discrete-state-space models for fast, accurate thermal analysis of 2.5D and 3D multi-chiplet architectures. The tool supports heterogeneous chiplet layouts, non-uniform thermal grids, anisotropic material properties, transient and steady-state simulations, and heatmap generation. The models are validated against FEM simulations and are designed to enable rapid thermal evaluation during architecture exploration.
https://github.com/AlishKanani/MFIT
mRI: Multi-modal 3D Human Pose Estimation Dataset
mRI is a multi-modal dataset for 3D human pose estimation using mmWave radar, RGB-D cameras, and inertial sensors. The dataset focuses on rehabilitation movements and home-based health monitoring applications, with over 5 million frames collected from 20 subjects. It supports research in human pose estimation, action detection, multi-modal learning, and privacy-aware sensing for health monitoring.
DS3: Domain-Specific System-on-Chip Simulator
DS3 is a system-level simulation framework for domain-specific systems-on-chip. It enables researchers to model heterogeneous SoC resources, task graphs, scheduling policies, dynamic power and thermal management, and workload execution behavior. DS3 has been used to study design and runtime management techniques for domain-specific architectures and energy-efficient heterogeneous computing systems.
https://github.com/segemena/DS3
OpenHealth Initiative
OpenHealth is an open-source wearable health monitoring platform that provides hardware and software components for autonomous collection of clinically relevant sensor data. The platform includes wearable device designs, standard software interfaces, and reference implementations for applications such as human activity recognition and gesture recognition. It supports research and education in wearable health, embedded sensing, low-power IoT systems, and health-focused machine learning.
The datasets are available on our GitHub page.
https://sites.google.com/view/openhealth-wearable-health/home
Multi-Objective Design Optimization for Flexible Hybrid Electronics
This release provides design-optimization tools for mapping applications onto flexible hybrid electronic systems that combine both flexible and rigid resources. The framework supports multi-objective exploration of system-level tradeoffs, including performance, energy, area, and implementation constraints, enabling automated design-space exploration for flexible hybrid electronics.
Rigid IC Placement for Flexible Hybrid Electronic Systems
This release provides algorithms for placing rigid integrated circuits within flexible hybrid electronic platforms. The tool supports physical design exploration by considering placement constraints, interconnect requirements, and system-level objectives, helping designers identify efficient layouts for systems that combine rigid silicon components with flexible substrates.