← Home

About SiliconBadgers

We’re a student chip-design club at UW–Madison. Learn digital logic, write and test hardware code, and work toward a chip design with other students. You can join without prior experience.

Our first project is an AI-inference accelerator that can run Qwen3.5, with plans to tape it out next year. More info coming soon...

From AI to Silicon

Follow a prompt through the system we’re building.

A prompt travels from a computer through a RISC-V CPU and the Badger Processing Unit The RISC-V CPU runs llama.cpp and offloads selected computations to the accelerator inside the Badger Processing Unit. The accelerator exchanges data with external memory within the same package, then returns results to the CPU. The CPU sends the response back to the computer. This is a conceptual illustration.

Inside the accelerator

Host Interface

Top-Level Controller

Memory Controller

SRAM

Compute Unit