Skip to main content

Upcoming Tutorial for Beethoven at ISCA 2026

🚀 Overview​

Welcome to the Beethoven Tutorial, your gateway to mastering open-source accelerator design! Beethoven lets you quickly prototype powerful accelerated systems based on the "separation of concerns" philosophy. Focus purely on crafting your accelerator Core, while Beethoven handles the heavy lifting by auto-generating optimized system components, making high-performance system creation both enjoyable and accessible.

🎓 What You Will Learn​

In this hands-on tutorial, you will master the entire lifecycle of accelerator development:

  • Build from Scratch: Learn to design a custom hardware accelerator core from the ground up using Beethoven's high-level abstractions.
  • Real-World Deployment: Go beyond RTL-level simulation: You will watch your design work end-to-end on a live AWS EC2 F2 instance.
  • Full-Stack Integration: Discover how hardware accelerators are seamlessly integrated into software stacks.
  • Measure Speedups: Witness firsthand how specialized hardware provides significant performance improvements for real-world workloads.

📅 Tutorial Details​

  • 🕒 Date: Saturday, June 27, 8:00 AM – 1:00 PM
  • 👥 Expected Audience: 10–30 participants. Researchers, industry professionals, and anyone interested in accelerator technologies are welcomed. No prior FPGA or Beethoven knowledge required!

🎤 Organizers​

  • Lisa Wu Wills (Duke University, lisa@cs.duke.edu)
  • Chris Kjellqvist (Duke University)
  • Mason Ma (Duke University)
  • Mansi Choudhary (Duke University)
  • Ning Liang (Duke University)

🗓️ Tutorial Schedule​

🕒 Duration📖 Topic
30 min🎵 Intro to Beethoven & Hardware Landscape
20 min🧩 Chisel Preface
30 min🧩 Code Structures and Abstractions
20 min🧩 Memory Streams
20 min🧩 Accelerator Configuration
20 min🧩 On-Chip Memory Abstractions
20 min🧩 Software Integration and Test Benches
30 min⚙️ Hands-on: Build a Simple Accelerator Core
50 min⚙️ Hands-on: Generate an Accelerated System
30 min📊 Hands-on: Performance & Power Evaluation

☁️ Infrastructure​

Participants will get hands-on experience with AWS EC2 F2 cloud instances generously funded by Duke University. Deploy and experiment with your Beethoven-generated systems in the cloud seamlessly!

👩‍🏫 Speaker Bios​

Lisa Wu Wills​

Assistant Professor of Computer Science and ECE at Duke University. Prior to Duke, she was a postdoctoral researcher at UC Berkeley and a research scientist at Intel Labs. Her research interests include computer architecture and microarchitecture, hardware acceleration, hardware-software co-design, and emerging applications in big data, healthcare, and artificial intelligence. Wills has a PhD in computer science from Columbia University. Her research is recognized via various awards such as an NSF CAREER Award, a VMware Early Career Faculty Grant, IEEE Micro Top Picks (x3) and Honorable Mentions (x2), and best paper awards from MICRO and ISPASS.

Chris Kjellqvist​

Chris is the lead architect of the Beethoven framework. His research leverages modern hardware description languages’ flexible, generative abilities and programming abstractions to provide scalable and reusable SoC infrastructure for hardware accelerator development.

Mason Ma​

A fifth-year PhD student in Computer Science at Duke University. His research focuses on efficient software and hardware design for privacy-preserving computing, with a particular emphasis on advancing fully homomorphic encryption (FHE) through optimizations in arithmetic, compilers, and hardware accelerators. He has developed hardware architectures and ML compilers that optimize FHE computations, achieving significant speedups and efficiency improvements for privacy-preserving natural language processing and data analysis tasks.

Mansi Choudhary​

A fourth-year PhD student in ECE at Duke University. Her primary area of research is computer architecture, with an emphasis on workload analysis, performance modeling, and hardware acceleration through architectural and microarchitectural enhancements for domain-specific applications, including AI. Her work aims to improve performance and power efficiency in these systems.

Ning Liang​

A second-year PhD student in Computer Science at Duke University. Ning's research investigates hardware and software optimizations for more efficient and accurate LLM serving systems. Specifically, his work focuses on accelerating vector databases and retrieval-augmented generation.