Fastcomp vs upstream performance

Hello, we've been running under the old fastcomp branch for a while with great performance, and I'm trying to switch over to upstream but running into a performance issue. We've found that upstream is performing ~3.5x *slower* in our use case and I can't explain what's happening.

One difference I've seen is that upstream seems to be repeatedly re-calling the same function whereas fastcomp only makes a single call. I can't figure out how to prevent this from happening.

Here's what we can see in fastcomp, note that we have single, uninterrupted calls:

However, you can see that upstream seems to be splitting the functions and weaving different calls:

For our fastcomp builds, we used the following parameters via Makefile:
`--bind -O2 --llvm-opts 2 -s FETCH=1 -s BUILD_AS_WORKER=1 -s ALLOW_MEMORY_GROWTH=1 -s ASSERTIONS=1 -s ABORTING_MALLOC=0 -s TOTAL_MEMORY=250MB -s BINARYEN_TRAP_MODE=clamp -std=c++11`

For upstream, we use the following parameters via CMakeLists:
`-lembind -O3 -flto=thin -std=c++11 -s WASM=1 -s FETCH=1 -s BUILD_AS_WORKER=1 -s ALLOW_MEMORY_GROWTH=1 -s ASSERTIONS=1 -s ABORTING_MALLOC=0 -s TOTAL_MEMORY=250MB`

Is there something I'm missing? I've been digging through the documentation and trying different combinations to try and close the gap, but haven't been successful yet. Any help is appreciated!