Hah! A few years ago I did something similar. I was in the process of porting some C code (a physics engine) to javascript. After porting the code I benchmarke

Hah! A few years ago I did something similar. I was in the process of porting some C code (a physics engine) to javascript. After porting the code I benchmarked it - and, no surprises - it was waay slower than the original C code. One reason C is faster than javascript is that C structs get "inlined" into the containing object. For example, in C struct Body { vec2 pos; vec2 velocity; } would just be 1 object. But the equivalent javascript code would allocate 3 objects instead. I inlined vec2 (replacing it with pos_x, pos_y, etc) and performance got a lot better. But I was curious what other structs were thrashing the garbage collector. So I added a call into all my constructors which generated a stack trace (new Error().stack), and used the stack trace as the key in a javascript object - with the value being the number of times that stack trace was seen. After sorting and printing the result, I had a hit list of the hottest stack traces which were thrashing V8's garbage collector. I fixed all the worst call sites and by the time I was done performance improved by about 3x or so!