Emscripten just orders by largest first. JS engines then typically parse linearly, and in firefox's case, it can also send off functions it parses in asm.js blocks directly to AOT compilation. I'm not sure if it does anything sophisticated there, I think it just sends them to the first available worker thread (or to a queue if not are free) as they are parsed, which is in source order.
- Alon On Thu, Jul 17, 2014 at 8:40 PM, Ryan Kelly <[email protected]> wrote: > On 18/07/2014 12:59 PM, Alon Zakai wrote: > > Very cool! > > > > Perhaps we should add this to tools/ ? > > If it turns out to be generally useful, sure! I will also play around > with the algorithm a little more and see if I can get it running a bit > faster. > > > Note btw that we already reorder functions, we put the biggest ones > > first (this helps reduce parallel AOT compilation times by reducing the > > chance of a core being idle). Might be worth keeping the order as close > > as possible to the original, while doing your optimization. > > Interesting. How do the functions get assigned to cores? Is it a > simple task queue populated in source order, or something more involved? > > > Ryan > > > > On Thu, Jul 17, 2014 at 7:42 PM, Ryan Kelly <[email protected] > > <mailto:[email protected]>> wrote: > > > > > > Hi All, > > > > > > I've just written up a little experiment I did to improve the > compressed > > file size of PyPy.js: > > > > https://rfk.id.au/blog/entry/cromulate-improve-compressibility/ > > > > The basic idea was to re-order the code for function definitions so > that > > similar functions are closer together, and hence likely to compress > > better. For such a naive trick it was surprisingly effective - > around a > > 10% decrease in compressed file size for my (admittedly very large > and > > bloated) use case. > > > > If anyone else feels like taking it for a spin, I'd be curious to > hear > > whether and how well it works on other emscriptenized codebases. > > > > > > Cheers, > > > > Ryan > > > > -- > > 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] > > <mailto:emscripten-discuss%[email protected]>. > > 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] > > <mailto:[email protected]>. > > 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]. > 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]. For more options, visit https://groups.google.com/d/optout.
