The incoming branch now has much faster dynamic linking throughput.
Earlier, we did each call between modules using a slow external call, now
we check if the target is in the same module, and do a fast asm.js call if
so.

This reduces the throughput overhead from a 100% or so slowdown (i.e.,
twice as slow), to just 5-10% slower. See more details in the docs,

https://github.com/kripken/emscripten/wiki/Linking

If you've been interested in dynamic linking but worried about the
overhead, hopefully 5-10% is small enough to make dynamic linking practical
for you. It's a good way to reduce compile times during development, for
example. Another use case is splitting up a large codebase in order to
reduce memory usage during startup, which is currently a problem on chrome
( https://github.com/WebAssembly/design/issues/108#issuecomment-120541752
), this was the motivation for me optimizing dynamic linking now.

- Alon

-- 
You received this message because you are subscribed to the Google Groups 
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to