Show HN: Autograd.c – A tiny ML framework built from scratch (opens in new tab)

At best you'd be restricted to the forward mode, which would still double stack pressure. If you needed reverse mode you'd need 2x stack, and the back sweep over the stack based tape would have the nearly perfectly unoptimal "grain". If you allows the higher order operators (both push out and pull back), you're going to end up with Jacobians & Hessians over nontrivial blocks. That's going to need the heap. It's still better than an unbounded loop tape, though.

We had all these issues back in 2006 when my group was implementing autograd for C++ and, later, a computer algebra system called Axiom. We knew it'd be ideal for NN; I was trying to build this out for my brother who was porting AI models to GPUs. (This did not work in 2006 for both HW & math reasons.)

> Is there a compiler-autograd "library"?

Do you mean the method theano is using? Anyway, the performance bottleneck often lies in matrix multiplication or 2D-CNN (which can be reduced to matmul). Compiler autograd wouldn't save much time.

We would need to mirror jax architecture more. Since the jax is sort of jit arch wise. Basically you somehow need a good way to convert computational graph to machine code while at compile time also perform a set of operations on the graph.

I believe Enzyme comes close to what you describe. It works on the LLVM IR level.

https://enzyme.mit.edu

a heap-free implementation could be a really cool direction to explore. thanks!

i think you might be interested in MLIR/IREE: https://github.com/openxla/iree

yeah, exactly. it's for explicit ownership transfer. you always own what you receive, sum it, release both inputs, done. no mutation tracking, no aliasing concerns.

https://github.com/sueszli/autograd.c/blob/main/src/autograd...

i wonder whether there is a more clever way to do this without sacrificing simplicity.