A lot like OpenCL
'River Trail' provides things like ParallelArray which has a method that can execute a user-given function in data-parallel (ie: running the function for each element in the ParallelArray, passing the function the index in the ParallelArray as well as any other params that the user initially passed through). So effectively SIMD and quite a bit like invoking on GPU kernels in OpenCL - in fact, Intel even says something to this effect.
Fractured web programming...
"I wonder if we are witnessing the beginnings of the fracturing of Web programming?"
They've open sourced a version so if its any good the only thing to stop it becoming a standard would be patents and standards organisations.
And who in the standards organisations are going to be against ECMAScript developing into a powerful too. Other than the obvious two...
You're confusing open-source and standards? Just because they opened the source code it doesn't make them any better than Google's Dart project. For me, for now, it's just another browser-based technology. In the future, if it becomes a standard, then we'll see. Also, more things would stop it from becoming a standard, it must be measured how much good it'd be for the Web and if it wouldn't favor Intel chips, and we don't have that information.
I guess the extensions just happen to be x86 specific? After all, who else has all those cores idly doing nothing?
Interestingly over on the IE blog, which is refreshingly detailed, there's a discussion of using "promises" for concurrent programming, the sister of parallel programming.
Running as an interpreted language on a multi-platform browser like Firefox, I don't see how they could be. If Intel can persuade ECMA to swallow this, you can be sure that *someone* will maintain the ARM and Power versions of the Firefox extension.
'I guess the extensions just happen to be x86 specific? After all, who else has all those cores idly doing nothing?'
This has nothing to do with cores, from what the article is describing this is vector processing - and that's something that ARM SoCs are perfectly capable of.
It is OpenCL specific.
So while it is not x86 specific, it apllies only to vector problems - stuff that can be parallelised using OpenCL. While there is a host of problems which fall into that category - image transform, video transform, etc they are only a small fraction of what can be made faster through parallelisation. It will do nothing with regards to accelerating "classic" software. For that you need threads (or something similar).
Like Web Workers then?
In the blue corner, it's Intel with River Trail and in the red corner, Dart from Google...
Meanwhile a C++ version would probably run 100 times faster.
C++ version may be 100 times faster
There may however be phantom technical problems raised in Standards Committees by certain parties to prevent you using your computer to its max from the browser.
Turing Tests of Great Minds ... for a Masterpiece of Network Servers
... in an AI TakeOver/SMART MakeOver
River Trail ..... Sounds Real Good Virtual Programming.
"we have made a 10X improvement on existing sites"
Please stop quoting that crap-trap. The reason they've madea 10X improvement is because it was so incredibly slow to start with. Even now their "per-PC optimised browser code" is far slower than native binaries