The Advanced Computing Lab
Created September 2024
The Advanced Computing Lab (ACL) in the AI Division at Carnegie Mellon University's Software Engineering Institute (SEI) has extensive expertise in optimizing software performance on diverse hardware architectures, and hardware and system design for software-based systems. The ACL applies this expertise to systems utilizing machine learning in the interests of national security.
What We Do
ACL capabilities and areas of expertise include the following:
- AI/ML accelerator benchmarking
- Algorithmic performance optimization
- Compiler design and assessment
- Data-intensive computing
- Edge computing and SWaP-C constraints
- Field Programmable Gate Array (FPGA) design and system integration
- Heterogenous computing
- High Performance Computing
- Homomorphic encryption
- Machine learning enhanced physics simulation
- Neuromorphic computing
- Physics-inspired neural networks
- Quantum computing and algorithms
- Software Defined Radio
- Sparse linear algebra and graphs
Sampling of Current and Recent Projects
Recent ACL projects of note include the following:
Hardware Software Co-design: The Co-design for Edge AI internal research and development funded project is targeting development of System on Chips (SoCs) targeted for specific applications. The technology focus is on high level synthesis for FPGAs. Learn more:
- Software-Hardware Codesign for Machine Learning Workloads (workshop)
- Co-Design for Edge AI: Application-Specific System on Chip (article)
- Research Review 2023: Co-design for Edge AI (video)
Portable High-Performance Inference on the Tactical Edge (PHITE): PHITE applies performance engineering processes to the analysis of existing open-source ML frameworks for embedded systems, to inform the development and optimization of a portable software library that can achieve significantly higher performance across a range of targeted embedded devices. Learn more:
- PHITE: Portable High-Performance Inference at the Tactical Edge (presentation)
- Portable High-Performance Inference on the Tactical Edge (PHITE) (article)
- PHITE: Improved Situational Awareness through AI at the Edge (fact sheet)
- Preview of PHITE: Portable High-Performance Inference at the Tactical Edge (video)
Spiral AI/ML: Enabling platform developers to realize high-performance AI/ML applications on leading-edge hardware architectures faster and cheaper. Learn more:
- Spiral/AIML: Resource-Constrained Co-Optimization for High-Performance, Data-Intensive Computing (article)
- Spiral AI/ML Collection (collection)
- Spiral/AIML: Frontiers of Graph Processing in Linear Algebra (poster)
Test and Evaluation for DARPA ERI programs: The ACL acts as an Expert First User for the government team on a multitude of DARPA ERI programs. Learn more:
Spotlight on ACL Members
Learn more about featured ACL researchers and read recent publications highlighted below.
Featured ACL Researchers
 |
 |
 |
 |
| Dr. John Wohlbier Principal Researcher ACL Lab Lead Author Profile |
Dr. Scott McMillan Principal Engineer Author Profile |
Dr. Jason Larkin Senior Researcher Author Profile |
Daniel Justice Software Developer Author Profile |
Recent Publications
- SMaLL: Software for Rapidly Instantiating Machine Learning Libraries (ACM)
- C++ and Interoperability Between Libraries: The GraphBLAS C++ Specification (IEEE)
- GraphBLAS: C++ Iterators for Sparse Matrices (IEEE)
- NWGraph: A Library of Generic Graph Algorithms and Data Structures in C++20 (Schloss Dagstuhl)
- Delayed Asynchronous Iterative Graph Algorithms (IEEE)
- Introduction to GraphBLAS 2.0 (IEEE)
- LAGraph: Linear Algebra, Network Analysis Libraries, and the Study of Graph Algorithms (IEEE)
- Unstructured-Grid CFD Algorithms on Many-Core Architectures (Research Gate)