My project in 2023

Prerequisite

“Hardware system design (4190.309A)” is a course in SNUCSE, covering hardware-software co-design: specifically custom accelerator for neural network.

Till 2022

Tech stack covered

  • MNIST classification (caffe2, C++)
  • Verilog (Vivado)
  • FPGA (Xilinx Zedboard)

What I did in 2023

Refactored to

  • focus on instruction set and compiler
  • be sustainable with open source

Tech stack covers

  • Image classification QAT (torch)
  • Simple compiler (python)
  • RTL simulator (Amaranth)