AESIP: Arch-aware ASIP-ISA Co-Design via Program Synthesis
A hardware-software co-optimization framework for efficient ASIP design using e-graphs, program synthesis, and equality saturation.
A hardware-software co-optimization framework for efficient ASIP design using e-graphs, program synthesis, and equality saturation.
A novel framework for designing composable, heterogeneous chiplet-based neural network accelerators with systematic design space exploration.
Linear programming formulation for simultaneous register and duplication logic minimization in superconducting circuits.
Domain-specific language and code generation framework for agile development of mixed binary-stochastic circuits.
Graph Neural Network-based approach for accurate and fast prediction of HLS metrics with feature extraction from reports.
Scalable recommendation system implementation using Hadoop cluster with Spark and MapReduce for distributed computing.
Custom database implementation with SQL-like syntax supporting multi-threading, thread pools, and parallel processing.
Kernel-level implementation of advanced scheduling algorithms including Earliest Deadline First and Lottery scheduling.