read a couple of books about compilers

LLVM Compiler for RISC-V Architecture

Describes details of risc-v vectorization support in llvm. It should be noted that the implementation of vector operations in risc-v was done later than in Intel and sve in arm64 - they took into account many flaws (like made explicit masks for vector operations) and were implemented in a much more convenient way from the programmer's point of view

On other hand any HW vendor can add it's own ISA subset and support of this custom processors in compiler can become very segmented and pure nightmare

 

Also I want to note that support of risc-v vectors in LLVM carefully avoids MLIR (IMHO second most overrated thing after LLM) - to do this they even had to patch their holy cow tablegen

 

Drawbacks:

• there is no introduction about LLVM IR/risc-v specific IR, so long IR listings are very hard to follow
• author don't give link to source code implementing some algo. Fortunately elixir indexed whole LLVM source tree
  

4/5

Dive into Deep Learning Compiler

As far as I know, this is the only book describing AI/ML compilers so far. Also TVM looks very promising - unlike monsters like XLA/iree it is compact and observable for mere mortals

Drawbacks:

• book is not completed - last two chapter about NN & deployment are just "place holder"
• it's unclear why for matrix multiplication on CUDA they didn't get cublas as base case
• and openblas for cpu version

Despite this, considering that the book is freely downloadable, my rating is 4 out of 5