Curious about the internals of dynamic dispatch

General Usage Performance
13

Hmm, I noticed that my types are of the same size so there is a concern of some optimization there. I changed B to also contain a Float64/f64.

Now we have Julia:

Dynamic Dispatch:
  4.673 ms (0 allocations: 0 bytes)
Static Dispatch
  256.708 μs (0 allocations: 0 bytes)

and Rust:

dynamic dispatch        time:   [4.4012 ms 4.4029 ms 4.4048 ms]
                        change: [+361.64% +362.24% +362.81%] (p = 0.00 < 0.05)
                        Performance has regressed.
Found 2 outliers among 100 measurements (2.00%)
  2 (2.00%) high severe

static dispatch         time:   [260.19 µs 260.35 µs 260.53 µs]
                        change: [+172.74% +173.19% +173.81%] (p = 0.00 < 0.05)
                        Performance has regressed.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) high mild
  6 (6.00%) high severe

Looks like they match up pretty closely now. The point of dictionary lookup vs array lookup is a good one. I suppose Julia is forced to be a little slower on that front to allow for extra flexibility

2 Likes