circt VS mlir-aie

MLIR (https://mlir.llvm.org/) is a quickly growing compiler toolkit which attempts to synthesize the learnings of LLVM and currently powers compilers for programming languages, machine learning and circuit design (https://github.com/llvm/circt). and there are a ton of companies with real employees working on it (including Microsoft) and MLIR is at the core of Chris Lattner’s new company, ModularAI. I’d recommend taking a look at it, there are a large number of ways to get involved and a number of paths from contributor to employee.

Might be good to target the CIRCT infrastructure at some point.

Even small optimizations like removing unused pins from internal modules are often times opposed.

Chris Lattner and others are currently working on an "industry" version of firrtl as part of the CIRCT hardware compiler framework: https://github.com/llvm/circt

Did you see the work being done on CIRCT? https://github.com/llvm/circt

I remember one of the reasons you did not want to use firrtl was that its compiler is implemented in Scala and thus hard to integrate into other projexts. CIRCT will solve that problem by providing a firrtl compiler implemented in C++. Other languages like Verilog/VHDL and new high level languages for HLS-like designs are also on the todo list.

Relevant: https://github.com/llvm/circt

>come up with a "better" (performant, cheaper, easier to use, etc.) solution than GPUs for ML applications

you probably are aware but Xilinx themselves is attempting this with their versal aie boards which (in spirit) similar to GPUs, in that they group together a programmable fabric of programmable SIMD type compute cores.

https://www.xilinx.com/support/documentation/architecture-ma...

i have not played with one but i've been told (by a xilinx person, so grain of salt) the flow from high-level representation to that arch is more open

https://github.com/Xilinx/mlir-aie