Why create a new standard when there is already asm.js?

… especially since pthreads (Mozilla pthreads, Chromium pthreads) and SIMD (simd.js, Chromium SIMD, simd.js in asm.js) are coming to JavaScript.

There are two main benefits WebAssembly provides:

  1. The kind of binary format being considered for WebAssembly can be natively decoded much faster than JavaScript can be parsed (experiments show more than 20× faster). On mobile, large compiled codes can easily take 20–40 seconds just to parse, so native decoding (especially when combined with other techniques like streaming for better-than-gzip compression) is critical to providing a good cold-load user experience.

  2. By avoiding the simultaneous asm.js constraints of AOT-compilability and good performance even on engines without specific asm.js optimizations, a new standard makes it much easier to add the features :unicorn: required to reach native levels of performance.

Of course, every new standard introduces new costs (maintenance, attack surface, code size) that must be offset by the benefits. WebAssembly minimizes costs by having a design that allows (though not requires) a browser to implement WebAssembly inside its existing JavaScript engine (thereby reusing the JavaScript engine’s existing compiler backend, ES6 module loading frontend, security sandboxing mechanisms and other supporting VM components). Thus, in cost, WebAssembly should be comparable to a big new JavaScript feature, not a fundamental extension to the browser model.

Comparing the two, even for engines which already optimize asm.js, the benefits outweigh the costs.

What are WebAssembly’s use cases?

WebAssembly was designed with a variety of use cases in mind.

Can WebAssembly be polyfilled?

We think so. There was an early prototype with demos [1, 2], which showed that decoding a binary WebAssembly-like format into asm.js can be efficient. And as the WebAssembly design has changed there have been more experiments with polyfilling.

Overall, optimism has been increasing for quick adoption of WebAssembly in browsers, which is great, but it has decreased the motivation to work on a polyfill.

It is also the case that polyfilling WebAssembly to asm.js is less urgent because of the existence of alternatives, for example, a reverse polyfill - compiling asm.js to WebAssembly - exists, and it allows shipping a single build that can run as either asm.js or WebAssembly. It is also possible to build a project into two parallel asm.js and WebAssembly builds by just flipping a switch in emscripten, which avoids polyfill time on the client entirely. A third option, for non-performant code, is to use a compiled WebAssembly interpreter such as binaryen.js.

However, a WebAssembly polyfill is still an interesting idea and should in principle be possible.

Is WebAssembly only for C/C++ programmers?

As explained in the high-level goals, to achieve a Minimum Viable Product, the initial focus is on C/C++.

However, by integrating with JavaScript at the ES6 Module interface, web developers don’t need to write C++ to take advantage of libraries that others have written; reusing a modular C++ library can be as simple as using a module from JavaScript.

Beyond the MVP, another high-level goal is to improve support for languages other than C/C++. This includes allowing WebAssembly code to allocate and access garbage-collected (JavaScript, DOM, Web API) objects :unicorn:. Even before GC support is added to WebAssembly, it is possible to compile a language’s VM to WebAssembly (assuming it’s written in portable C/C++) and this has already been demonstrated (1, 2, 3). However, “compile the VM” strategies increase the size of distributed code, lose browser devtools integration, can have cross-language cycle-collection problems and miss optimizations that require integration with the browser.

Which compilers can I use to build WebAssembly programs?

WebAssembly initially focuses on C/C++, and a new, clean WebAssembly backend is being developed in upstream clang/LLVM, which can then be used by LLVM-based projects like Emscripten and PNaCl.

As WebAssembly evolves it will support more languages than C/C++, and we hope that other compilers will support it as well, even for the C/C++ language, for example GCC. The WebAssembly working group found it easier to start with LLVM support because they had more experience with that toolchain from their Emscripten and PNaCl work.

We hope that proprietary compilers also gain WebAssembly support, but we’ll let vendors speak about their own platforms.

The WebAssembly Community Group would be delighted to collaborate with more compiler vendors, take their input into consideration in WebAssembly itself, and work with them on ABI matters.

Will WebAssembly support View Source on the Web?

Yes! WebAssembly defines a text format to be rendered when developers view the source of a WebAssembly module in any developer tool. Also, a specific goal of the text format is to allow developers to write WebAssembly modules by hand for testing, experimenting, optimizing, learning and teaching purposes. In fact, by dropping all the coercions required by asm.js validation, the WebAssembly text format should be much more natural to read and write than asm.js. Outside the browser, command-line and online tools that convert between text and binary will also be made readily available. Lastly, a scalable form of source maps is also being considered as part of the WebAssembly tooling story.

What’s the story for Emscripten users?

Existing Emscripten users will get the option to build their projects to WebAssembly, by flipping a flag. Initially, Emscripten’s asm.js output would be converted to WebAssembly, but eventually Emscripten would use WebAssembly throughout the pipeline. This painless transition is enabled by the high-level goal that WebAssembly integrate well with the Web platform (including allowing synchronous calls into and out of JavaScript) which makes WebAssembly compatible with Emscripten’s current asm.js compilation model.

Is WebAssembly trying to replace JavaScript?

No! WebAssembly is designed to be a complement to, not replacement of, JavaScript. While WebAssembly will, over time, allow many languages to be compiled to the Web, JavaScript has an incredible amount of momentum and will remain the single, privileged (as described above) dynamic language of the Web. Furthermore, it is expected that JavaScript and WebAssembly will be used together in a number of configurations:

  • Whole, compiled C++ apps that leverage JavaScript to glue things together.
  • HTML/CSS/JavaScript UI around a main WebAssembly-controlled center canvas, allowing developers to leverage the power of web frameworks to build accessible, web-native-feeling experiences.
  • Mostly HTML/CSS/JavaScript app with a few high-performance WebAssembly modules (e.g., graphing, simulation, image/sound/video processing, visualization, animation, compression, etc., examples which we can already see in asm.js today) allowing developers to reuse popular WebAssembly libraries just like JavaScript libraries today.
  • When WebAssembly gains the ability to access garbage-collected objects :unicorn:, those objects will be shared with JavaScript, and not live in a walled-off world of their own.

Why not just use LLVM bitcode as a binary format?

The LLVM compiler infrastructure has a lot of attractive qualities: it has an existing intermediate representation (LLVM IR) and binary encoding format (bitcode). It has code generation backends targeting many architectures and is actively developed and maintained by a large community. In fact PNaCl already uses LLVM as a basis for its binary format. However, the goals and requirements that LLVM was designed to meet are subtly mismatched with those of WebAssembly.

WebAssembly has several requirements and goals for its Instruction Set Architecture (ISA) and binary encoding:

  • Portability: The ISA must be the same for every machine architecture.
  • Stability: The ISA and binary encoding must not change over time (or change only in ways that can be kept backward-compatible).
  • Small encoding: The representation of a program should be as small as possible for transmission over the Internet.
  • Fast decoding: The binary format should be fast to decompress and decode for fast startup of programs.
  • Fast compiling: The ISA should be fast to compile (and suitable for either AOT- or JIT-compilation) for fast startup of programs.
  • Minimal nondeterminism: The behavior of programs should be as predictable and deterministic as possible (and should be the same on every architecture, a stronger form of the portability requirement stated above).

LLVM IR is meant to make compiler optimizations easy to implement, and to represent the constructs and semantics required by C, C++, and other languages on a large variety of operating systems and architectures. This means that by default the IR is not portable (the same program has different representations for different architectures) or stable (it changes over time as optimization and language requirements change). It has representations for a huge variety of information that is useful for implementing mid-level compiler optimizations but is not useful for code generation (but which represents a large surface area for codegen implementers to deal with). It also has undefined behavior (largely similar to that of C and C++) which makes some classes of optimization feasible or more powerful, but can lead to unpredictable behavior at runtime. LLVM’s binary format (bitcode) was designed for temporary on-disk serialization of the IR for link-time optimization, and not for stability or compressibility (although it does have some features for both of those).

None of these problems are insurmountable. For example PNaCl defines a small portable subset of the IR with reduced undefined behavior, and a stable version of the bitcode encoding. It also employs several techniques to improve startup performance. However, each customization, workaround, and special solution means less benefit from the common infrastructure. We believe that by taking our experience with LLVM and designing an IR and binary encoding for our goals and requirements, we can do much better than adapting a system designed for other purposes.

Note that this discussion applies to use of LLVM IR as a standardized format. LLVM’s clang frontend and midlevel optimizers can still be used to generate WebAssembly code from C and C++, and will use LLVM IR in their implementation similarly to how PNaCl and Emscripten do today.

Why is there no fast-math mode with relaxed floating point semantics?

Optimizing compilers commonly have fast-math flags which permit the compiler to relax the rules around floating point in order to optimize more aggressively. This can include assuming that NaNs or infinities don’t occur, ignoring the difference between negative zero and positive zero, making algebraic manipulations which change how rounding is performed or when overflow might occur, or replacing operators with approximations that are cheaper to compute.

These optimizations effectively introduce nondeterminism; it isn’t possible to determine how the code will behave without knowing the specific choices made by the optimizer. This often isn’t a serious problem in native code scenarios, because all the nondeterminism is resolved by the time native code is produced. Since most hardware doesn’t have floating point nondeterminism, developers have an opportunity to test the generated code, and then count on it behaving consistently for all users thereafter.

WebAssembly implementations run on the user side, so there is no opportunity for developers to test the final behavior of the code. Nondeterminism at this level could cause distributed WebAssembly programs to behave differently in different implementations, or change over time. WebAssembly does have some nondeterminism in cases where the tradeoffs warrant it, but fast-math flags are not believed to be important enough:

  • Many of the important fast-math optimizations happen in the mid-level optimizer of a compiler, before WebAssembly code is emitted. For example, loop vectorization that depends on floating point reassociation can still be done at this level if the user applies the appropriate fast-math flags, so WebAssembly programs can still enjoy these benefits. As another example, compilers can replace floating point division with floating point multiplication by a reciprocal in WebAssembly programs just as they do for other platforms.
  • Mid-level compiler optimizations may also be augmented by implementing them in a JIT library in WebAssembly. This would allow them to perform optimizations that benefit from having information about the target and information about the source program semantics such as fast-math flags at the same time. For example, if SIMD types wider than 128-bit are added, it’s expected that there would be feature tests allowing WebAssembly code to determine which SIMD types to use on a given platform.
  • When WebAssembly adds an FMA operator :unicorn:, folding multiply and add sequences into FMA operators will be possible.
  • WebAssembly doesn’t include its own math functions like sin, cos, exp, pow, and so on. WebAssembly’s strategy for such functions is to allow them to be implemented as library routines in WebAssembly itself (note that x86’s sin and cos instructions are slow and imprecise and are generally avoided these days anyway). Users wishing to use faster and less precise math functions on WebAssembly can simply select a math library implementation which does so.
  • Most of the individual floating point operators that WebAssembly does have already map to individual fast instructions in hardware. Telling add, sub, or mul they don’t have to worry about NaN for example doesn’t make them any faster, because NaN is handled quickly and transparently in hardware on all modern platforms.
  • WebAssembly has no floating point traps, status register, dynamic rounding modes, or signalling NaNs, so optimizations that depend on the absence of these features are all safe.

What about mmap?

The mmap syscall has many useful features. While these are all packed into one overloaded syscall in POSIX, WebAssembly unpacks this functionality into multiple operators:

  • the MVP starts with the ability to grow linear memory via a grow_memory operator;
  • proposed future features :unicorn: would allow the application to change the protection and mappings for pages in the contiguous range 0 to memory_size.

A significant feature of mmap that is missing from the above list is the ability to allocate disjoint virtual address ranges. The reasoning for this omission is:

  • The above functionality is sufficient to allow a user-level libc to implement full, compatible mmap with what appears to be noncontiguous memory allocation (but, under the hood is just coordinated use of memory_resize and mprotect/map_file/map_shmem/madvise).
  • The benefit of allowing noncontiguous virtual address allocation would be if it allowed the engine to interleave a WebAssembly module’s linear memory with other memory allocations in the same process (in order to mitigate virtual address space fragmentation). There are two problems with this:

    • This interleaving with unrelated allocations does not currently admit efficient security checks to prevent one module from corrupting data outside its heap (see discussion in #285).
    • This interleaving would require making allocation nondeterministic and nondeterminism is something that WebAssembly generally tries to avoid.

Why have wasm32 and wasm64, instead of just an abstract size_t?

The amount of linear memory needed to hold an abstract size_t would then also need to be determined by an abstraction, and then partitioning the linear memory address space into segments for different purposes would be more complex. The size of each segment would depend on how many size_t-sized objects are stored in it. This is theoretically doable, but it would add complexity and there would be more work to do at application startup time.

Also, allowing applications to statically know the pointer size can allow them to be optimized more aggressively. Optimizers can better fold and simplify integer expressions when they have full knowledge of the bitwidth. And, knowing memory sizes and layouts for various types allows one to know how many trailing zeros there are in various pointer types.

Also, C and C++ deeply conflict with the concept of an abstract size_t. Constructs like sizeof are required to be fully evaluated in the front-end of the compiler because they can participate in type checking. And even before that, it’s common to have predefined macros which indicate pointer sizes, allowing code to be specialized for pointer sizes at the very earliest stages of compilation. Once specializations are made, information is lost, scuttling attempts to introduce abstractions.

And finally, it’s still possible to add an abstract size_t in the future if the need arises and practicalities permit it.

Why have wasm32 and wasm64, instead of just using 8 bytes for storing pointers?

A great number of applications don’t ever need as much as 4 GiB of memory. Forcing all these applications to use 8 bytes for every pointer they store would significantly increase the amount of memory they require, and decrease their effective utilization of important hardware resources such as cache and memory bandwidth.

The motivations and performance effects here should be essentially the same as those that motivated the development of the x32 ABI for Linux.

Even Knuth found it worthwhile to give us his opinion on this issue at some point, a flame about 64-bit pointers.

Will I be able to access proprietary platform APIs (e.g. Android / iOS)?

Yes but it will depend on the WebAssembly embedder. Inside a browser you’ll get access to the same HTML5 and other browser-specific APIs which are also accessible through regular JavaScript. However, if a wasm VM is provided as an “app execution platform” by a specific vendor, it might provide access to proprietary platform-specific APIs of e.g. Android / iOS.