Post #1484998

I'm working on the hash table optimizations I will demonstrate in my talk. I've focused on higher-level changes thus far (mostly related to memory accesses), but I started dipping my toes into some fun micro-optimizations yesterday. I tried to simplify some hot branch conditions, but the performance didn't improve as much as I expected. So I examined the generated machine code ... but perf annotate is refusing to show me the source lines corresponding to the disassembled instructions. I ended up constructing a behemoth of an objdump invocation to help me analyze the code, and I discovered some options I didn't know about before: objdump --disassemble --no-show-raw-insn --visualize-jumps=extended-color --disassembler-color=extended --demangle=rust --source --line-numbers --inlines $path | less You can also make Cargo invoke this for you, via --config "target.'cfg(all())'.runner = "..."; see https://doc.rust-lang.org/stable/cargo/reference/config.html#targettriplerunner. #programming #rust #optimization