Re:Why Is C Faster Than Java (2009)
by josephg on Monday December 27, 2021 @04:18PM
(#29706879)
> The same algorithm not going to run faster in Rust than tuned C, period.
Weirdly enough, before I added that optimization my rust code was ~20% faster than the C code anyway. And I have no idea why - the programs were (as far as I could tell) identical. I was using the same compiler backend (llvm) and before alias analysis was turned on for rust. And -march=native in both cases.
Could just be a weird coincidence of the compiler’s inlining decisions - though I suspect not. I tried investigating it but I don’t understand x86_64 assembly enough to understand how the binaries differ.
Weirdly enough, before I added that optimization my rust code was ~20% faster than the C code anyway. And I have no idea why - the programs were (as far as I could tell) identical. I was using the same compiler backend (llvm) and before alias analysis was turned on for rust. And -march=native in both cases.
Could just be a weird coincidence of the compiler’s inlining decisions - though I suspect not. I tried investigating it but I don’t understand x86_64 assembly enough to understand how the binaries differ.