Do you know when the next milestone checkpoint is going to be? Code size has always been a sticking point for us. I'd like to run a few experiments with release build optimizations enabled so that I can send around a demo link (bonus points if it works in multiple browsers). I've heard size reductions can be between 33-44% and I want to get a good idea of where my 27MB app is going to land.
Floh, thank you for being an awesome vanguard! I bet those issues would have bitten us in the field if they had made it to release. On Tuesday, August 16, 2016 at 3:19:41 AM UTC-4, Floh wrote: > > Main problem for someone playing around with WebAssembly at the moment is > IMHO that the required tools and publicly available web browsers are > getting out of sync every few weeks(?). You need the emscripten SDK, > binaryen, and a Spidermonkey JS build (I use the one from mozilla-central, > and compile myself), and there must be a browser which supports the current > version of emitted WebAssembly (usually the latest Nightly or Chrome Canary > build). > > Sometimes the generated code isn't accepted by the current Nightly or > Canary. Other then that, all my stuff was running out of the box with no > code changes required. > > On Chrome Canary I've got a feeling that the JS engine is getting a bit > fragile in the past few months for asm.js, at least my 8-bit emulator demo > is pretty good at triggering bugs in Canary recently ;) I'm not sure if > this is related to the current work going on in WebAssembly (I've had cases > where WebAssembly was running stable, but the asm.js version triggered > aw-snap pages). I'm just hoping these problem don't make it into the live > version... > > Here's the current ticket: > https://bugs.chromium.org/p/chromium/issues/detail?id=633497 > > And here's a recent one: > https://bugs.chromium.org/p/chromium/issues/detail?id=611976 > > Cheers, > -Floh. > > Am Montag, 15. August 2016 23:26:28 UTC+2 schrieb Robert Goulet: >> >> What's the status as of today? >> >> On Thursday, February 18, 2016 at 10:19:31 PM UTC-5, awt wrote: >>> >>> Thanks for your reply Alon. Looking to try out wasm soon :) >>> >>> On Thursday, February 18, 2016 at 3:14:08 AM UTC+8, Alon Zakai wrote: >>>> >>>> Still too early. The spec isn't final and browsers don't support it >>>> yet. Hopefully over the next few months, though. >>>> >>>> Emscripten+Binaryen support for emitting wasm was done early, so that >>>> we know it's all ready for when browsers are. Also to help test browsers >>>> by >>>> emitting content for them to try on. >>>> >>>> Performance, however, will not change much. asm.js in most browsers >>>> today is already running close to native speed, minus some sandboxing, so >>>> we can't expect a big change there. However, some things will help, like >>>> hardware min/max, load/store with offset. But I would guess less than 5% >>>> speedup in throughput. (On the other hand, asm.js will get SIMD and >>>> threads >>>> before wasm, so there will be a period where it is faster.) >>>> >>>> Code size should be noticeably smaller. At least 33%. And startup >>>> should be far faster as well due to avoiding parsing JS and going straight >>>> from binary to codegen (10x or more for the parse stage). >>>> >>>> On Wed, Feb 17, 2016 at 7:10 AM, awt <[email protected]> wrote: >>>> >>>>> Hi, >>>>> >>>>> I understand that Emscripten can now emit WebAssembly thru Binaryen >>>>> but is the generated WebAssembly supported on Chrome or Firefox? >>>>> >>>>> Do we also have any benchmarks on the performance as well as code >>>>> size? Thanks. >>>>> >>>>> -- >>>>> 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.
