[webkit-dev] how to do JIT benchmark

[x yz]

I used to see that calling convention change lead to certain % of efficiency improvement. in general how we can do a speed benchmark?
rgds
joe