Wasmer 1.0 Released As The Universal WebAssembly Runtime Outside The Browser

[phoronix]

Phoronix: Wasmer 1.0 Released As The Universal WebAssembly Runtime Outside The Browser

As we have been expecting in recent weeks, Wasmer 1.0 has been released as the "universal WebAssembly runtime" for helping to accelerate WASM adoption and new use-cases outside of the web browser...

http://www.phoronix.com/scan.php?page=news_item&px=Wasmer-1.0-Released

[uid313]

So when are they building a CPU that executes WASM?
And how does WASM compare against JVM, .NET CLR, and RISC-V?

[onlyLinuxLuvUBack]

It says everything, but I don't see holyc-compiler -> WA, or watcomc -> WA, or lazarus-> WA... where they at ?

[Yttrium]

how does WASM compare against JVM, .NET CLR, and RISC-V?

Really bad or good depending on what you want. WASM is focused on producing simple dense binaries. The instruction set is more minimal than Risk-v by a longshot, it only contains 4 primitive types ( i32,i64,f32,f64) I was sceptical at first but WASM is thought trough and does not inherit the same flaws as flash or JS. Though you can still write js trough their JS api

[Viki Ai]

So when are they building a CPU that executes WASM?
And how does WASM compare against JVM, .NET CLR, and RISC-V?

Might be a while waiting on a hard-implementation since my reading around is that JITing to a decent RISC (ARM64, PPC, RISC-V) architecture can produce better performance than a [conventionally simple, relatively] hard CPU (which is effectively interpreting at clock-speed, without access to much context for optimisation, which seems to be the main reason JIT trupms interpreting, and apparently strait-out compiling in some cases!). I'm sure some really really clever person or team could achieve the advantages of JIT in a hardware-based interpreter with some clever look-ahead-stuff or similar (effectively some sort of hardware psuedo-JIT), but the research in that area isn't even particularly advanced for existing CPUs (I suspect, ironically, the x86 architecture may have the most going on in this area as modern x86 chips are effectively already doing this, though with a far more convoluted bytestream than WebAssembly).
...
Where is Transmetta when you need them?!

[Ladis]

So when are they building a CPU that executes WASM?

I don't know about WASM, but Apple flaunts that M1 is the fastest CPU for JS. They say it contains hardware optimizations for thar, similar to some CPU having optimizations for Java (SPARC?).

[zboszor]

I guess the first letter is turned upside down, making it a twisted MASM. It was MASM all the way up from the DOS times.

[jonkoops]

So when are they building a CPU that executes WASM?

I don't think anyone will be building a CPU that uses WASM instructions. WASM is made as an intermediate representation that can be easily converted into machine code in a sandboxed environment.

[Michael_S]

So when are they building a CPU that executes WASM?
And how does WASM compare against JVM, .NET CLR, and RISC-V?

I think the killer features, if there are any, are:

1. Cross-architecture, cross-programming language FFI. A Python library compiled to WASM on Windows x86_64 can, at runtime, call functions from a C++ library compiled to WASM on macOS ARM. And that runtime can be running on some RISC-V device.

2. Code you can use natively with Wasmer and Wasmtime and also inside a browser.

3. Lots of existing applications can be recompiled into WASM.

I think WASM is interesting because I think it could, in theory, break the walled gardens. If the apps a person likes on their iPhone and the games they liked from Windows 10 can be compiled to WASM, then they can move to any operating system that has WASM.

[Artim]

I think WASM is interesting because I think it could, in theory, break the walled gardens. If the apps a person likes on their iPhone and the games they liked from Windows 10 can be compiled to WASM, then they can move to any operating system that has WASM.

In theory. But fact is especially companies like Apple and Microsoft will probably never ever do that. Sure, MS might move some of their applications to wasm, probably especially the ones that already run on the web, but they probably won't let you run it in something like Wasmer. And Apple will probably do anything to port iOS Apps to wasm. And it remeains to be seen how much wasm Apple will allow in Safari. They want developers to make proper iOS Apps, and in the first place they want the total control over any app running on iOS. Or what do you think why they copied the concept of Android's instant apps and make them available via NFC- or QR-Codes for cases the easiest way would have been something like a PWA?
I mean it would be great if you could finally leave the walled gardens and use any app you want on any platform you want, but in the end some OS might not survive without some exclusive apps that in the best case everyone wants to use. Users would abandon ship if they have no compelling reason to stay (other than habit). And there are things wasm will not be able to fix if the root of the problem lies to deep, like missing hardware support.