DMC projects have produced a number of open source software packages that are available on GitHub.

Converged Applications

Causal Inference and ML Methods for MIP Analysis of Security Constrained Unit (SCY₀) Commitment

• SCUC: Simulation and security constraint evaluation code
• Y₀: Causal inference engine

Heterogenous Computing

Transfer Learning for Building Energy Modeling (TransBeam)

• TransBeam: Transfer learning for building energy modeling

Hybrid Advanced Workflows (HAW)

• MCL: Minos Computing Library (MCL) is a modern task-based, asynchronous programming model for extremely heterogeneous systems
• SHAD: Scalable High-performance Algorithms and Data-structures (SHAD) C++ library
• SHADES: The SHAD Exploration System

Software-Defined Architecture for Data Analysis (SO(DA)²)

• ARENA: Asynchronous reconfigurable accelerator ring architecture serving as a selective solution on what the future HPC cluster will be like
• Bambu: A free framework for the high-level synthesis of complex applications
• Energy Characterization of Graph Workloads Data: Energy characterization study of six graph algorithm benchmark suite kernels with a variety of input graphs (9 synthetic, 3 real-world) on a dual-socket x86-based system
• OpenCGRA: Open-source framework for Coarse-Grained Reconfigurable Arrays (CGRAs)
• SODA-OPT: Enabling System Level Design in MLIR

• SODA Toolchain Production Docker Image: The SODA toolchain leverages mlir to extract, optimize, and translate high-level code snippets into LLVM IR, so that they can be synthesized by our high-level synthesis tool of choice

Computational-Flow Architecture (CFA): Designing Non-Von-Neumann Architecture for Future Data-Centric Computing

• ARENA: Asynchronous reconfigurable accelerator ring architecture serving as a selective solution on what the future HPC cluster will be like
• OpenCGRA: Open-source framework for Coarse-Grained Reconfigurable Arrays (CGRAs)
• TCBNN: Tensor-Core Accelerated Binarized Neural Network (TCBNN) is an efficient Binarized-Neural-Network (BNN) design accelerated by NVIDIA Turing Bit-Tensor-Cores

Application-Algorithm-Architecture Co-Design for Large-Scale, Sparse Tensor/Matrix Methods (HiParTI)

• HiPartTI: Hierarchical Parallel Tensor Infrastructure (HiParTI) supports fast essential sparse tensor operations and tensor decompositions on multicore CPU and GPU architectures

Fixing amdahl's law within the limits of accelerated systems (fallacy)

• MemGaze: MemGaze is a memory analysis toolset that combines high-resolution trace analysis and low overhead measurement, both with respect to time and space

A compiler infrastructure for data-model convergence (duomo)

• COMET: Domain specific COMpiler for Extreme Targets