One thing to note is that when doing such a test, make sure to do the test before loading up the Emscripten page, since the Emscripten polyfill at https://github.com/kripken/emscripten/blob/master/src/ecmascript_simd.js will be added to the page, which implements the polyfilled window.SIMD object.
2016-05-19 15:32 GMT+03:00 Robert Goulet <[email protected]>: > Thanks! > > On Thursday, May 19, 2016 at 5:26:44 AM UTC-4, Karsten Schmidt wrote: >> >> Hi Robert, >> >> The experimental SIMD reference pages are here: >> https://developer.mozilla.org/en-US/docs/Web/JavaScript/SIMD_types >> >> https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/SIMD >> >> To check for native support, just verify if window.SIMD is defined... >> >> Hth! >> >> On 18 May 2016 at 21:40, Robert Goulet <[email protected]> wrote: >> > Depending on how fast is the SIMD adoption in browsers we might just >> not >> > ship a non-SIMD version, having two separate builds is something we >> want to >> > avoid ideally. >> > >> > I know its coming along very well in Firefox 49, but what about other >> > browsers? Does anyone have any insights about SIMD.js support in other >> > browsers? >> > >> > And how does one test for SIMD support? Is this documented? >> > >> > Thanks >> > >> > On Wednesday, May 18, 2016 at 12:58:05 PM UTC-4, jj wrote: >> >> >> >> Cool, great to hear! We do currently provide the SIMD polyfill as a >> >> fallback, so the SIMD capable page should run on non-SIMD browsers as >> well. >> >> Although, unfortunately the SIMD portions will run like 100x slower, >> since >> >> the polyfill emulation is doing constant FFI operations between JS and >> >> asm.js sides. Improving the emulation to be faster is something that >> was >> >> concluded would require way too difficult machinery to pull off in >> asm.js. >> >> This was discussed in >> https://github.com/kripken/emscripten/issues/4090 and >> >> https://github.com/kripken/emscripten/issues/3783. >> >> >> >> For final releases that need to support non-SIMD browsers as well, we >> >> recommend you build two versions of the page and do a feature test in >> the >> >> .html to serve the appropriate build. Sorry to say that we don't have >> >> anything better to offer here. When discussing wasm simd support a few >> >> months back with the spec developers, we did raise this use case, so >> >> hopefully we'll have a better fallback machinery to provide for wasm. >> >> >> >> 2016-05-18 19:32 GMT+03:00 Robert Goulet <[email protected]>: >> >>> >> >>> Been starting to use SIMD (SSE3) in our project and in Firefox >> Nightly >> >>> (49) its now providing a substancial performance improvement, good >> job! >> >>> >> >>> However, not all browsers support it, so what is the plan regarding >> this? >> >>> Is Emscripten going to provide some function to test if browser >> supports >> >>> SIMD or not, so that we can branch in our code? Or is there a better >> way to >> >>> deal with this? >> >>> >> >>> 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. >> >> >> >> -- >> Karsten Schmidt >> http://postspectacular.com | http://thi.ng | http://toxiclibs.org >> > -- > 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.
