SIMD For JavaScript Continues Coming Along

[phoronix]

Phoronix: SIMD For JavaScript Continues Coming Along

SIMD for JavaScript continues to be worked on by Mozilla, Google, Intel, and others for better accelerating particular workloads in the web...

http://www.phoronix.com/vr.php?view=MTgyNjQ

[mark45]

SIMD for JS is like DRI3 - it doesn't remove the problem (X11) but makes it bearable.
The other shitty workaround around the shitty JavaScript was asm.js.
The other shitty workaround around JavaScript was adding float arrays for webgl as built in types because the devs discovered (omg!) that in the real world JavaScript sucks and needs to deal with types.
Who knows what other shitty workarounds they'll need to figure out to further work around the main problem - JavaScript.

[Luke_Wolf]

SIMD for JS is like DRI3 - it doesn't remove the problem (X11) but makes it bearable.
The other shitty workaround around the shitty JavaScript was asm.js.
The other shitty workaround around JavaScript was adding float arrays for webgl as built in types because the devs discovered (omg!) that in the real world JavaScript sucks and needs to deal with types.
Who knows what other shitty workarounds they'll need to figure out to further work around the main problem - JavaScript.

Well, the real problem is that people are trying to turn the web into emacs, and are missing vital points like: If you need SIMD... you probably shouldn't be writing it for the web client, and if you really must do so that you should really be using PNaCl instead.

[peppercats]

Well, the real problem is that people are trying to turn the web into emacs, and are missing vital points like: If you need SIMD... you probably shouldn't be writing it for the web client, and if you really must do so that you should really be using PNaCl instead.

Shame Firefox decided to not support pnacl, it really is better than asm.is.

Edit:just found out about pepper.js, going to check it out later. http://trypepperjs.appspot.com/

[Szzz]

SIMD for JS is like DRI3 - it doesn't remove the problem (X11) but makes it bearable.
The other shitty workaround around the shitty JavaScript was asm.js.
The other shitty workaround around JavaScript was adding float arrays for webgl as built in types because the devs discovered (omg!) that in the real world JavaScript sucks and needs to deal with types.
Who knows what other shitty workarounds they'll need to figure out to further work around the main problem - JavaScript.

I couldn't have said it any better.

Well, the real problem is that people are trying to turn the web into emacs, and are missing vital points like: If you need SIMD... you probably shouldn't be writing it for the web client, and if you really must do so that you should really be using PNaCl instead.

What you say is nonsense. If you need to use PNaCl instead of JS that means that something is wrong with JS.

[Luke_Wolf]

What you say is nonsense. If you need to use PNaCl instead of JS that means that something is wrong with JS.

Or maybe it just means that there's something wrong with using a domain specific language for general purpose programming. Javascript is okay but not wonderful for the task it was designed to do, which is to provide basic interactivity to web pages, it does a lot of boneheaded things because of the limitations that the w3c forces on it and it's interpreted dynamic nature, but if you're just using it for its proper task it works. The problem is that you have all of these people who want to take it outside of it's domain and try to use it as a general purpose programming language which it's not and wasn't designed to be.

[Szzz]

which it's not and wasn't designed to be.

Yes and this is its problem. There is nothing wrong in trying to add more cool things to webpages. What stops (or at least hinders) from doing this is js. That's why we need something better for web programming.

[Luke_Wolf]

Yes and this is its problem. There is nothing wrong in trying to add more cool things to webpages. What stops (or at least hinders) from doing this is js. That's why we need something better for web programming.

And here's where we're going to fundamentally disagree. In my opinion there IS something fundamentally wrong with trying to turn the web into an OS, many things actually. If you're exceeding what JS can do comfortably, you should really be writing an application for the desktop and using the general internet instead of trying to force it to work on the web.

[nanonyme]

And here's where we're going to fundamentally disagree. In my opinion there IS something fundamentally wrong with trying to turn the web into an OS, many things actually. If you're exceeding what JS can do comfortably, you should really be writing an application for the desktop and using the general internet instead of trying to force it to work on the web.

Not to forget many native desktop applications embed browser frames with html+css+js for gui that has nothing to do with the Internet

[Szzz]

And here's where we're going to fundamentally disagree. In my opinion there IS something fundamentally wrong with trying to turn the web into an OS, many things actually. If you're exceeding what JS can do comfortably, you should really be writing an application for the desktop and using the general internet instead of trying to force it to work on the web.

But you are proposinng to use PNaCl, isn't it fundamentally wrong?