This project is meant to be a DL acceleration framework that allows someone to build operations on top of it.
The idea is that LLVM has many impressive optimizations already (via its regular optimizer or Polly)
and so we should be trying to make it easy for them to do their jobs. In this spirit, we create llvm::Module
s and
run optimizations and compilation against those.
This project is very much a work in progress, and since everything I know about LLVM comes from reading the docs and seeing what produces (in my mind) correct IR through trial and error, I welcome feedback and PRs.
Requires:
llvm >= 6.0
On macOS simply run brew install llvm