Bingo. I've written high perf F# for a DB indexing and searching. The entire time I was wishing for allocation-free inlined closures and stack allocation. And for a few places, I'd really like an easy way to do asm or get really top notch codegen (integer decoding). Rust seems a lot like a fast ML. Not quite as concise as I'd like but worth it for the perf without being ugly mentally.
And the perf will surpass C in some situations due to abstraction. One popular open source platform spends 30% CPU time on malloc and strcpy because tracking ownership was so difficult and it wasn't obvious it'd be a hotspot. In Rust that would be a non issue from beginning.
You may be interested in MLton, which is an ML compiler that achieves high levels of optimisation by aggressively specialising your code: generic functions get specialised to the type and higher order functions get inlined to eliminate closure allocation.
And the perf will surpass C in some situations due to abstraction. One popular open source platform spends 30% CPU time on malloc and strcpy because tracking ownership was so difficult and it wasn't obvious it'd be a hotspot. In Rust that would be a non issue from beginning.