Hmm yeah I was worried that might happen. We’re going to have to do some science to figure out where it is actually beneficial to use the bulk memory instructions. Alternatively we can just pester engine implementers to optimize them better.
> On Jul 12, 2019, at 09:38, Lilit Darbinyan <[email protected]> wrote: > > oops typo - it's twice as slow WITH the flag. > >> On Friday, July 12, 2019 at 5:37:52 PM UTC+1, Lilit Darbinyan wrote: >> Thanks! >> >> Interestingly it's twice as slow without the flag. It's now 40ms instead of >> the 20ms. To summarize: >> >> Fastcomp 80ms >> LLVM 20ms >> LLVM with -mbulk-memory 40ms >> >> The profiler now shows this as the hot path instead of memmove: >> >> std::__2::vector<double, std::__2::allocator<double> >> >::insert(std::__2::__wrap_iter<double const*>, unsigned long, double const&) >> > > -- > 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]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/emscripten-discuss/7cf8a87b-bf54-49ea-96a1-cb337c903546%40googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- 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]. To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/654F8E88-D6B6-4085-98FA-F4AFD3B0AF71%40google.com. For more options, visit https://groups.google.com/d/optout.
