How close can pure functional programming get to C?

A developer optimized a DEFLATE compression library written in Roc, a functional programming language focused on runtime performance, and achieved performance within 8.9% of libdeflate (a highly-optimized C implementation) on AMD Ryzen processors and just 2.6% slower on M1 MacBook Pro. This demonstrates that pure functional programming can get remarkably close to C's performance for compute-intensive tasks, even though libdeflate still maintains an edge due to architecture-specific SIMD optimizations that are difficult to implement in higher-level languages.