This repository contains an Emscripten system library for utilizing WebGPU from a C/C++ codebase, along with a few small C code examples on how to use it.
To utilize the library in your own application, copy the contents of the lib/ directory into your project:
- lib/lib_webgpu.h
- lib/lib_webgpu.js
- lib/lib_webgpu.cpp
- lib/lib_webgpu_fwd.h
- lib/lib_webgpu_cpp20.cpp or lib/lib_webgpu_cpp11.cpp, depending on if your compiler has C++20 or C++11.
Then #include "lib_webgpu.h" to access the API, compile in lib_webpgu.cpp and lib_webgpu_cpp20.cpp with the rest of your project files, and finally link with --js-library /absolute/path/to/lib_webgpu.js on the Emscripten command line to include the code. See the provided CMakeLists.txt for example usage.
For your convenience, a forward declaration header is also provided, and can be included with #include "lib_webgpu_fwd.h".
It is also possible to target WebGPU outside the browser via Dawn. When doing so, also compile the dawn-specific file with your project:
TODO: add more instructions about targeting Dawn natively.
The repository was last updated to match the API IDL of the WebGPU specification as of 🗓 6th of September 2024.
If you want to get to building WebGPU content quickly, try the following:
- As a prerequisite, download Emscripten if you don't have it already:
> git clone https://github.com/emscripten-core/emsdk
> cd emsdk
> emsdk install latest ninja-git-release-64bit
> emsdk activate latest ninja-git-release-64bit
> source ./emsdk_env.sh # Linux and macOS
> emsdk_env # Windows
Whenever you open a new command prompt, first run the emsdk_env line in emsdk/ root directory.
- In an emsdk_env-enabled command prompt, clone and build wasm_webgpu samples:
> git clone https://github.com/juj/wasm_webgpu
> cd wasm_webgpu
> mkdir build
> cd build
> emcmake cmake -G Ninja ../samples
> ninja
- Launch the
emrunad hoc web server to host the built .html pages:
> emrun .
An ad hoc web server index will pop up, allowing you to launch the different sample programs in this repository.
This bindings library is developed with the following:
For the most parts, the JavaScript side WebGPU API is directly mapped 1:1 over to WebAssembly side to enable developers to write WebGPU code in C/C++ by using the official specification IDL as reference.
Type names and structs follow a naming convention WGpu*, mapped from JS names by transforming GPUAdapter -> WGpuAdapter. API function names use a prefix wgpu_*, and are mapped using the convention GPUCanvasContext.configure(...) -> wgpu_canvas_context_configure(canvasContext, ...). Enums and #defines use a prefix WGPU_, e.g. GPUPowerPreference -> WGPU_POWER_PREFERENCE.
A few exceptions to this are done in the name of accommodating better Wasm<->JS language marshalling, noted where present in the lib_webgpu.h header.
If you are pondering whether to use this repository or the WebGPU support header provided in the Emscripten repository, this 1:1 API mapping with JS point is the main difference between the two interfaces. The Emscripten WebGPU header allows targeting WebGPU by using the Dawn WebGPU C/C++ API as a reference, whereas this repository allows targeting WebGPU via the JavaScript Browser API as a reference.
The primary design goal of the library is to provide absolutely best runtime speed and minimal generated code size overhead, carefully shaving down every individual byte possible. The intent is to enable using this library in extremely code size constrained deployment scenarios.
The library is implemented very C-like, void of high-level JavaScript abstractions, and manually tuned to produce smallest code possible. Past experience developing language bindings has taught that this kind of strategy works best to provide the thinnest JS<->Wasm language marshalling layer that does not get in the way as "bloaty".
In order to achieve the smallest code size, Closure Compiler should be used. Wasm_webgpu is fully Closure compatible. To enable Closure minification, copy the Closure externs file into your project:
and specify the Emscripten linker arguments --closure 1 and --closure-args=--externs=/path/to/webgpu-closure-externs.js when linking your project.
Another design goal is to minimize the amount of JS temporary garbage that is generated. Unlike WebGL, WebGPU API is unfortunately quite trashy, and it is not possible to operate WebGPU without generating some runaway garbage each rendered frame. However, the binding layer itself minimizes the amount of generated garbage as much as possible.
Some WebGPU features do not interop well between JS and Wasm if translated 1:1. Buffer mapping is one of these features. To help JS<->Wasm interop, this library provides custom functions wgpu_buffer_read_mapped_range() and wgpu_buffer_write_mapped_range() that do not exist in the official WebGPU specification.
For an example of how this works in practice, see the sample vertex_buffer/vertex_buffer.c
To enable easy uploading of image URLs to WebGPU textures, an extension function wgpu_load_image_bitmap_from_url_async() is provided. For an example of this, see the sample texture/texture.c
When building with Emscripten linker flag -sJSPI (requires Emscripten 3.1.59 or newer), the following extra functions are available:
navigator_gpu_request_adapter_syncandnavigator_gpu_request_adapter_sync_simple: Synchronously request a GPUAdapter.wgpu_adapter_request_device_syncandwgpu_adapter_request_device_sync_simple: Synchronously request a GPUDevice.wgpu_buffer_map_sync: Synchronously map a GPUBuffer.wgpu_present_all_rendering_and_wait_for_next_animation_frame: Presents current rendered frame, runs browser event loop, and waits until next animation frame. See the sample clear_screen/clear_screen_sync.c for an example.
These functions enable a synchronous variant of the _async functions offered in the WebGPU specification. These can be useful for prototyping and test suites etc., though note that there are some concerns over the runtime performance of JSPI, so be very mindful about profiling the impact on performance when using JSPI.
Additionally, note that the web browser will keep pumping web events while WebAssembly execution is suspended via JSPI. Therefore you may see your web event callbacks being fired at odd times. This can break ordering and re-entrancy semantics of your expected code execution. To remedy this problem, you can try guarding/queueing your event callbacks whenever a JSPI suspend is currently in effect. See the function wgpu_sync_operations_pending() for this.
It is possible to perform WebGPU rendering from a dedicated background Worker thread using the Emscripten Wasm Worker, pthreads or proxy-to-pthread abstractions.
The following API functions are provided to manage OffscreenCanvas rendering:
void offscreen_canvas_create(OffscreenCanvasId id, int width, int height);
void canvas_transfer_control_to_offscreen(const char *canvasSelector, OffscreenCanvasId id);
void offscreen_canvas_post_to_worker(OffscreenCanvasId id, emscripten_wasm_worker_t worker);
void offscreen_canvas_post_to_pthread(OffscreenCanvasId id, pthread_t pthread);
WGPU_BOOL offscreen_canvas_is_valid(OffscreenCanvasId id);
void offscreen_canvas_destroy(OffscreenCanvasId id);
int offscreen_canvas_width(OffscreenCanvasId id);
int offscreen_canvas_height(OffscreenCanvasId id);
void offscreen_canvas_size(OffscreenCanvasId id, int *outWidth, int *outHeight);
void offscreen_canvas_set_size(OffscreenCanvasId id, int width, int height);See lib_webgpu.h header file for detailed documentation, and the samples/offscreen_canvas/ subdirectory for code snippets.
When targeting OffscreenCanvas with Wasm Workers, pass the Emscripten compiler flag -sWASM_WORKERS for each compilation unit, and the linker flags -sWASM_WORKERS -sENVIRONMENT=web,worker for the final link. There is no need to pass the Emscripten -sOFFSCREENCANVAS_SUPPORT or -sOFFSCREENCANVASES_TO_PTHREAD= linker flags in this mode (doing so will just increase generated code size for no gain)
When targeting OffscreenCanvas with pthreads, pass the Emscripten compiler flag -pthread for each compilation unit, and the linker flags -pthread -sENVIRONMENT=web,worker for the final link. Likewise, in this mode there is no need to specify the Emscripten -sOFFSCREENCANVAS_SUPPORT and -sOFFSCREENCANVASES_TO_PTHREAD= linker flags.
Finally, when targeting OffscreenCanvas with the proxy-to-pthread option, pass the Emscripten compiler flag -pthread for each compilation unit, and the linker flags -pthread -sENVIRONMENT=web,worker -sOFFSCREENCANVAS_SUPPORT -sOFFSCREENCANVASES_TO_PTHREAD=#canvas -lGL -sPROXY_TO_PTHREAD for the final link. Note that the linkage to the WebGL support library is needed here for historical reasons. This might change in the future.
Wasm_Webgpu supports each of the three main memory models that Emscripten supports:
- 2GB mode: link with
-sMAXIMUM_MEMORY=2GBor less, - 4GB mode: link with
-sMAXIMUM_MEMORY=4GBor less, - Wasm64 mode: link with
-sMEMORY64=1 -sMAXIMUM_MEMORY=16GBor some other value > 4GB.
When building code samples, pass -DMEMORY64=1 to CMake to test compiling in Wasm64 build mode.
Wasm_Webgpu requires Emscripten 3.1.35 or newer.
Run test.py --browser="C:\Users\clb\AppData\Local\Google\Chrome SxS\Application\chrome.exe" <test1> <test2> ... <testN> to run unit tests.
Replace the cmdline to --browser= with location of your own WebGPU supporting browser, or omit to run in system default browser.
In <testN> you can pass names of tests to run. Test names are substring matches to filter filenames inside test/ subdirectory, so for example test.py adapter device would run all tests with substring adapter or device in it. Do not specify any test names to run through all tests in the suite.
To add a new test in the suite, create a new .cpp file with the test contents in the test/ subdirectory, and run test.py name_of_cpp_file to run the test.
Several test cases are available in the samples/ subdirectory.
Don't expect flashy demos. The test cases exercise features relevant to data marshalling between WebAssembly and JavaScript languages, and are not intended to showcase fancy graphical effects.
To build the samples, first install Emscripten via Emsdk, then enter Emsdk command line environment (emsdk_env), and type
cd path/to/wasm_webgpu
mkdir build
cd build
emcmake cmake ../samples -DCMAKE_BUILD_TYPE=Debug # Or MinSizeRel, RelWithDebInfo or Release
make -jOn Windows, the last make command is not available, so either install Mingw32-make via emsdk and run mingw32-make -j, or install Ninja via emsdk, then pass -G Ninja to the emcmake command line, and then run ninja instead of make.
For the smallest Clear Screen "hello world" example, see clear_screen.c/clear_screen.c.
There is also an Emscripten JSPI-enabled variant of the same demo, at clear_screen.c/clear_screen_sync.c.
The demo failing_shader_compilation/failing_shader_compilation.c tests handling of shader compilation errors.
The demo gpu_oom/gpu_oom.c exhausts the GPU VRAM, testing handling of GPU OOM events.
The demo hello_triangle/hello_triangle_minimal.c contains the smallest triangle rendering demo.
The variant hello_triangle/hello_triangle_verbose.c offers the same, but with verbose debug logging.
The demo offscreen_canvas/offscreen_canvas.c shows how to perform WebGPU rendering using OffscreenCanvas from a Wasm Worker.
If you are using pthreads, the variant offscreen_canvas/offscreen_canvas_pthread.c illustrates OffscreenCanvas rendering by using a pthread instead.
Finally, if you are using pthreads with the Emscripten -sPROXY_TO_PTHREAD build option, then check out the offscreen_canvas/offscreen_canvas_proxy_to_pthread.c code sample.
The sample texture/texture.c tests the wgpu_load_image_bitmap_from_url_async() API.
The test vertex_buffer/vertex_buffer.c shows an example of how to map a GPU buffer and use the function wgpu_buffer_write_mapped_range().