Comment by pbiggar
13 hours ago
When I think of how a "compiler" could make these optimizations, I think the right place is an optimizing LLM (so, just a regular coding agent that you prompted to find optimizations like this one), making the changes in source at the request of the developer. That provides the dev with adequate input on whether they would like to opt-in to an unsafe optimization like this one. The compiler can continue to do deterministically-safe optimizations.
What I'd like to enable this use of LLMs more recklessly is a compiler with formal methods that lets me guarantee equivalence between the opaque optimized code and something actually understandable.
Equivalence on what metrics? In theory what you are asking for makes sense, but I think it is very hard to actually specify what equivalent means in the context of an optimization process that needs to emit code with observably different behavior. Sometimes (although admittedly rarely) speeding up sections of code can even be undesirable for example branchless code for constant time algorithms that avoid timing or energy side channel leaks, or the much more mundane elimination of signed overflow checks or other undefined behavior quirks.
We'll get there eventually whether we like it or not. Just as soon as the current crop of AI/ML engineers retire, or .. get replaced with bots.
That seems like a horrible idea:
1. Do you really want the rust compiler to run at the speed of an llm?
2. Compiler optimisations are already extremely unpredictable with deterministic compilers[1], I hate to think how unpredictable your compiler would be.
3. What if someone else wants to build the software, do they have to decide on optimisations now? What if the optimisation depends on your features not available on old generations of CPU? (There is a reason we don’t compile with -march=native)
4. Compilers already have “unsafe” optimisations, but people rarely enable them (-ffast-math)
[1]: https://faultlore.com/blah/oops-that-was-important/
> Do you really want the rust compiler to run at the speed of an llm?
That... might actually be an improvement?
What do you mean "might"? I can crank Fable or Sol up to max intelligence and they'll spend an hour reviewing my rust SDK for working with our ADLSgen2's, and it'll still be done before the rust compiler has compiled the same project.
You misunderstand me. I'm saying that the developers can make these optimizations with LLMs, at the source level, and thus they don't need to be added to compilers.
Like just open Claude Code and ask it to find optimizations. That's the right place for this kind of optimization.