Open, Modular, Deep Learning Accelerator

Clone this repo:
  1. c18a224 Chisel Pipelined GEMM (#30) by Abhijit Davare · 5 weeks ago main
  2. 1954ff5 add unittest (#29) by Luis Vega · 7 weeks ago
  3. d5e8117 Chisel Pipelined ALU (#27) by Abhijit Davare · 7 weeks ago
  4. 74f23ff add scalastyle to test (#28) by Luis Vega · 7 weeks ago
  5. dfe9f57 [App][Verilator] change scalar add to parametric-vector-add (#26) by Luis Vega · 3 months ago

VTA Hardware Design Stack

Build Status

VTA (versatile tensor accelerator) is an open-source deep learning accelerator complemented with an end-to-end TVM-based compiler stack.

The key features of VTA include:

  • Generic, modular, open-source hardware
    • Streamlined workflow to deploy to FPGAs.
    • Simulator support to prototype compilation passes on regular workstations.
  • Driver and JIT runtime for both simulator and FPGA hardware back-end.
  • End-to-end TVM stack integration
    • Direct optimization and deployment of models from deep learning frameworks via TVM.
    • Customized and extensible TVM compiler back-end.
    • Flexible RPC support to ease deployment, and program FPGAs with the convenience of Python.