Instead of doing a naive O(n^2) liveness detection loop, use a bitmap
for values allocated to registers.
This cuts down validating time from 20% to 1.4% of runtime on the same
game as last commit.
This, along with moving the sources and destination out of the config
object, makes it so we don't have to double-deref to get to them on each
instruction, leading to a ~15% perf improvement on dispatch.
Namely, find an upper bound at validation time so we can allocate the
space when entering the frame.
Also drop labels at once instead of popping them off one at a time now
that we're using a Vector.
This still passes the values on the stack, but registers are now allowed
to cross a call boundary.
This is a very significant (>50%) improvement on the small call
microbenchmarks on my machine.
This commit adds a register allocator, with 8 available "register"
slots.
In testing with various random blobs, this moves anywhere from 30% to
74% of value accesses into predefined slots, and is about a ~20% perf
increase end-to-end.
To actually make this usable, a few structural changes were also made:
- we no longer do one instruction per interpret call
- trapping is an (unlikely) exit condition
- the label and frame stacks are replaced with linked lists with a huge
node cache size, as we only need to touch the last element and
push/pop is very frequent.
The average wasm function rarely goes over these bounds for the labels
(32 nested control structures), and 8 frames is just enough to clear
most initialization code/start section without allocating anything.
