10mb wasm file size?

[Wandalen]

I have a trouble trying to minimize wasm build of webgpu for my project. But still getting way more to make WebGPU compet-able with old JS WebGL engines.

I am wondering is any way to compile WebGPU shaders during compile-time and drop Naga from the build of WebGPU as it is major contributor to size of wasm file. Any chance to use WebGPU in a commercial project?

Does WebGPU have no_std option?

[kvark]

Hi!

I am wondering is any way to compile WebGPU shaders during compile-time

This has been requested on multiple occasions. WebGL case in particular sounds interesting, since WebGL is one big platform, which itself will provide safety. I filed #2280 to track this.

drop Naga from the build of WebGPU as it is major contributor to size of wasm file

Could you provide some numbers to see what is actually contributing to wasm size?
In particular, I'd be interested to know if the validation code contributes a lot.

Any chance to use WebGPU in a commercial project?

Sure, why not?

Does WebGPU have no_std option?

Please note this is wgpu library, not WebGPU the standard.
wgpu uses the standard library rather heavily - there is validation and resource tracking code. We can't have no_std.

[grovesNL]

The size will also change significantly depending on whether you're targeting the WebGPU backend, the WebGL backend, or both.

[Wandalen]

Guys, thank you for answers. That's helpful!

@grovesNL Which feature should I on/off to drop WebGPU backend? Isn't WebGPU off by default?

[Wandalen]

Could you provide some numbers to see what is actually contributing to wasm size?

Yes. I will.

[Wandalen]

There are too many things at the top of the list related to naga.
Not enough for rigorous proof of my assumption, but enough to claim it's very likely that Naga is a major contributor to the size of the web binary.

> twiggy top -n 30 input.wasm

 Shallow Bytes │ Shallow % │ Item
───────────────┼───────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
        303391 ┊    15.27% ┊ "function names" subsection
        200769 ┊    10.11% ┊ data[0]
         38903 ┊     1.96% ┊ naga::front::wgsl::Parser::parse_function_call_inner::h0936716646aba678
         23773 ┊     1.20% ┊ naga::front::wgsl::Parser::parse_statement::heed965287d7d609b
         21478 ┊     1.08% ┊ wgpu_core::command::render::<impl wgpu_core::hub::Global<G>>::command_encoder_run_render_pass_impl::h9605c202f99e941e
         20682 ┊     1.04% ┊ naga::front::wgsl::Parser::parse::h6e07072bcb57d1f6
         17823 ┊     0.90% ┊ naga::valid::expression::<impl naga::valid::Validator>::validate_expression::ha5ca28a871c281ff
         16909 ┊     0.85% ┊ naga::back::glsl::Writer<W>::write_expr::hfb26caef53b8934b
         14455 ┊     0.73% ┊ wgpu_core::device::queue::<impl wgpu_core::hub::Global<G>>::queue_submit::h61a4e027583d6d2d
         14171 ┊     0.71% ┊ naga::front::wgsl::Error::as_parse_error::h84f087ad4a215083
         13979 ┊     0.70% ┊ <wgpu_core::hub::Global<G> as core::ops::drop::Drop>::drop::hf479ec36e7222384
         13951 ┊     0.70% ┊ <wgpu_core::hub::Global<G> as core::ops::drop::Drop>::drop::h939417fcbf6c7f59
         12711 ┊     0.64% ┊ naga::back::glsl::Writer<W>::write_stmt::h9f47d80fbefadae6
         11654 ┊     0.59% ┊ naga::proc::typifier::ResolveContext::resolve::h39ddfec451c24e66
         11507 ┊     0.58% ┊ naga::proc::typifier::ResolveContext::resolve::hbfbe85bde9e415c4
         11027 ┊     0.56% ┊ naga::front::wgsl::Parser::parse_unary_expression::h8bc9730cfde25793
         10060 ┊     0.51% ┊ naga::front::wgsl::Parser::parse_type_decl_impl::hfa67146cccd620b7
          9375 ┊     0.47% ┊ naga::valid::Validator::validate::hffd62bc7d976bdb8
          8981 ┊     0.45% ┊ wgpu_core::device::life::LifetimeTracker<A>::triage_suspected::h9db1ff08ca10c1b3
          8235 ┊     0.41% ┊ naga::valid::function::<impl naga::valid::Validator>::validate_block_impl::h4730107b472d0835
          7994 ┊     0.40% ┊ codespan_reporting::term::renderer::Renderer::render_snippet_source::h16a5df15060ec533
          7862 ┊     0.40% ┊ wgpu_core::device::Device<A>::create_render_pipeline::h76b413737ebad1c8
          7630 ┊     0.38% ┊ wgpu_hal::gles::queue::<impl wgpu_hal::Queue<wgpu_hal::gles::Api> for wgpu_hal::gles::Queue>::submit::h4ae4840d4e08d149
          7496 ┊     0.38% ┊ codespan_reporting::term::views::RichDiagnostic<FileId>::render::hb25fec69a1bf5c44
          6780 ┊     0.34% ┊ wgpu_hal::gles::adapter::<impl wgpu_hal::gles::Adapter>::expose::h9a68520745fef021
          6686 ┊     0.34% ┊ wgpu_hal::gles::device::<impl wgpu_hal::Device<wgpu_hal::gles::Api> for wgpu_hal::gles::Device>::create_render_pipeline::h32bbd42f6b98a22c
          6648 ┊     0.33% ┊ naga::back::glsl::Writer<W>::write::he6e8effb5e8693cf
          6489 ┊     0.33% ┊ core::num::flt2dec::strategy::dragon::format_shortest::h6d64f97815ec3dae
          5904 ┊     0.30% ┊ wgpu_core::validation::Interface::check_stage::hbd6bb57505e1ff1f
          5898 ┊     0.30% ┊ naga::valid::analyzer::FunctionInfo::process_block::h5ed5585dff68dd7b
       1133311 ┊    57.05% ┊ ... and 3835 more.
       1986532 ┊   100.00% ┊ Σ [3865 Total Rows]