asmbench
A benchmark toolkit for assembly instructions using the LLVM JIT.
Usage
To benchmark latency and throughput of a 64bit integer add use the following command:
python -m asmbench 'add {src:i64:r}, {srcdst:i64:r}'
To benchmark two instructions interleaved use this:
python -m asmbench 'add {src:i64:r}, {srcdst:i64:r}' 'sub {src:i64:r}, {srcdst:i64:r}'
To find out more add -h for help and -v for verbose mode.