Home

Welcome to the AMD-Xilinx center of excellence at the University of Illinois, Urbana-Champaign.

The center is directed by CSL Professor Deming Chen, the Abel Bliss Professor of Engineering, and is home to the UIUC Heterogenous Adaptive Compute Cluster.

The AMD Center of Excellence and AMD HACC at UIUC investigate and propose system solutions for high-performance computing (HPC), distributed and heterogeneous computing, machine learning (ML), and other computation-intensive applications. This includes system integration and optimization, application acceleration, security, programming models, compilation, resource management, scheduling, and design automation, targeting AMD hardware systems, such as AMD FPGAs (e.g., Alveo and Versal devices), GPUs (e.g., MI200), SmartNICs (e.g., SN1000), and CPUs (e.g., Ryzen). The goal is to create new opportunities in data center and HPC domains by increasing the flexibility of the memory and storage hierarchy, improving design productivity and quality through advanced compilers and HLS (high-level synthesis), supporting coordinated and coherent computing across different hardware platforms, and exploiting parallelism at a large scale.

Concretely, we have following research thrusts:

  • Compilers and languages – programmability
    • New high-level synthesis solutions, programming models, and fast compilers.
  • Systems solutions – flexibility, scalability and heterogeneity
    • Innovative frameworks for unified memory, virtualization, scheduling and resource allocation.
    • Targeting multi-FPGA and multi-accelerator setups.
  • Distributed computing, networking, and storage – parallelization, security and reliability
    • Smart NICs, cloud RPC, NVMe over fabric, near-data acceleration, security and reliability.
    • P4 and eBPF targeting cloud computing.
  • Applications and algorithms acceleration – performance and efficiency
    • Tuning, optimizing, and accelerating HPC, ML, bioinformatics, encryption, and other types of applications.
  • Inter-XACC collaboration opportunities
    • E.g., unified memory work with Coyote of ETH HACC and graph processing with NUS HACC.