Wgpu now supports querying shader compilation info.
This allows you to get more structured information about compilation errors, warnings and info:
...
let lighting_shader = ctx.device.create_shader_module(include_wgsl!("lighting.wgsl"));
let compilation_info = lighting_shader.get_compilation_info().await;
for message in compilation_info
.messages
.iter()
.filter(|m| m.message_type == wgpu::CompilationMessageType::Error)
{
let line = message.location.map(|l| l.line_number).unwrap_or(1);
println!("Compile error at line {line}");
}By @stefnotch in #5410
- Avoid introducing spurious features for optional dependencies. By @bjorn3 in #5691
- Fix regression on OpenGL (EGL) where non-sRGB still used sRGB #5642
Wgpu supports now pipeline-overridable constants
This allows you to define constants in wgsl like this:
override some_factor: f32 = 42.1337; // Specifies a default of 42.1337 if it's not set.And then set them at runtime like so on your pipeline consuming this shader:
// ...
fragment: Some(wgpu::FragmentState {
compilation_options: wgpu::PipelineCompilationOptions {
constants: &[("some_factor".to_owned(), 0.1234)].into(), // Sets `some_factor` to 0.1234.
..Default::default()
},
// ...
}),
// ...By @teoxoy & @jimblandy in #5500
Due to a specification change write_timestamp is no longer supported on WebGPU.
wgpu::CommandEncoder::write_timestamp requires now the new wgpu::Features::TIMESTAMP_QUERY_INSIDE_ENCODERS feature which is available on all native backends but not on WebGPU.
By @wumpf in #5188
Many numeric built-ins have had a constant evaluation implementation added for them, which allows them to be used in a const context:
abs, acos, acosh, asin, asinh, atan, atanh, cos, cosh, round, saturate, sin, sinh, sqrt, step, tan, tanh, ceil, countLeadingZeros, countOneBits, countTrailingZeros, degrees, exp, exp2, floor, fract, fma, inverseSqrt, log, log2, max, min, radians, reverseBits, sign, trunc
By @ErichDonGubler in #4879, #5098
The following subgroup operations are available in wgsl now:
subgroupBallot, subgroupAll, subgroupAny, subgroupAdd, subgroupMul, subgroupMin, subgroupMax, subgroupAnd, subgroupOr, subgroupXor, subgroupExclusiveAdd, subgroupExclusiveMul, subgroupInclusiveAdd, subgroupInclusiveMul, subgroupBroadcastFirst, subgroupBroadcast, subgroupShuffle, subgroupShuffleDown, subgroupShuffleUp, subgroupShuffleXor
Availability is governed by the following feature flags:
wgpu::Features::SUBGROUPfor all operations exceptsubgroupBarrierin fragment & compute, supported on Vulkan, DX12 and Metal.wgpu::Features::SUBGROUP_VERTEX, for all operations exceptsubgroupBarriergeneral operations in vertex shaders, supported on Vulkanwgpu::Features::SUBGROUP_BARRIER, for support of thesubgroupBarrieroperation, supported on Vulkan & Metal
Note that there currently some differences between wgpu's native-only implementation and the open WebGPU proposal.
By @exrook and @lichtso in #5301
wgpu::Features::SHADER_INT64 enables 64 bit integer signed and unsigned integer variables in wgsl (i64 and u64 respectively).
Supported on Vulkan, DX12 (requires DXC) and Metal (with MSL 2.3+ support).
By @atlv24 and @cwfitzgerald in #5154
- Implemented the
Unorm10_10_10_2VertexFormat by @McMackety in #5477 wgpu-types'straceandreplayfeatures have been replaced by theserdefeature. By @KirmesBude in #5149wgpu-core'sserial-passfeature has been removed. Useserdeinstead. By @KirmesBude in #5149- Added
InstanceFlags::GPU_BASED_VALIDATION, which enables GPU-based validation for shaders. This is currently only supported on the DX12 and Vulkan backends; other platforms ignore this flag, for now. By @ErichDonGubler in #5146, #5046.- When set, this flag implies
InstanceFlags::VALIDATION. - This has been added to the set of flags set by
InstanceFlags::advanced_debugging. Since the overhead is potentially very large, the flag is not enabled by default in debug builds when usingInstanceFlags::from_build_config. - As with other instance flags, this flag can be changed in calls to
InstanceFlags::with_envwith the newWGPU_GPU_BASED_VALIDATIONenvironment variable.
- When set, this flag implies
wgpu::Instancecan now report whichwgpu::Backendsare available based on the build configuration. By @wumpf #5167-wgpu::Instance::any_backend_feature_enabled() +!wgpu::Instance::enabled_backend_features().is_empty()
- Breaking change:
wgpu_core::pipeline::ProgrammableStageDescriptoris now optional. By @ErichDonGubler in #5305. Features::downlevel{_webgl2,}_featureswas made const by @MultisampledNight in #5343- Breaking change:
wgpu_core::pipeline::ShaderErrorhas been moved tonaga. By @stefnotch in #5410 - More as_hal methods and improvements by @JMS55 in #5452
- Added
wgpu::CommandEncoder::as_hal_mut - Added
wgpu::TextureView::as_hal wgpu::Texture::as_halnow returns a user-defined type to match the other as_hal functions
- Added
- Allow user to select which MSL version to use via
--metal-versionwith Naga CLI. By @pcleavelin in #5392 - Support
arrayLengthfor runtime-sized arrays inside binding arrays (for WGSL input and SPIR-V output). By @kvark in #5428 - Added
--shader-stageand--input-kindoptions to naga-cli for specifying vertex/fragment/compute shaders, and frontend. by @ratmice in #5411 - Added a
create_validatorfunction to wgpu_coreDeviceto create nagaValidators. By @atlv24 #5606
- Implement the
device_set_device_lost_callbackmethod forContextWebGpu. By @suti in #5438 - Add support for storage texture access modes
ReadOnlyandReadWrite. By @JolifantoBambla in #5434
- Log an error when GLES texture format heuristics fail. By @PolyMeilex in #5266
- Cache the sample count to keep
get_texture_format_featurescheap. By @Dinnerbone in #5346 - Mark
DEPTH32FLOAT_STENCIL8as supported in GLES. By @Dinnerbone in #5370 - Desktop GL now also supports
TEXTURE_COMPRESSION_ETC2. By @Valaphee in #5568 - Don't create a program for shader-clearing if that workaround isn't required. By @Dinnerbone in #5348.
- OpenGL will now be preferred over OpenGL ES on EGL, making it consistent with WGL. By @valaphee in #5482
- Fill out
driveranddriver_info, with the OpenGL flavor and version, similar to Vulkan. By @valaphee in #5482
- Metal 3.0 and 3.1 detection. By @atlv24 in #5497
- Shader Model 6.1-6.7 detection. By @atlv24 in #5498
- Simplify and speed up the allocation of internal IDs. By @nical in #5229
- Use memory pooling for UsageScopes to avoid frequent large allocations. by @robtfm in #5414
- Eager release of GPU resources comes from device.trackers. By @bradwerth in #5075
- Support disabling zero-initialization of workgroup local memory in compute shaders. By @DJMcNab in #5508
- Improved
wgpu_haldocumentation. By @jimblandy in #5516, #5524, #5562, #5563, #5566, #5617, #5618 - Add mention of primitive restart in the description of
PrimitiveState::strip_index_format. By @cpsdqs in #5350 - Document and tweak precise behaviour of
SourceLocation. By @stefnotch in #5386 and #5410 - Give short example of WGSL
push_constantsyntax. By @waywardmonkeys in #5393 - Fix incorrect documentation of
Limits::max_compute_workgroup_storage_sizedefault value. By @atlv24 in #5601
- Fix
serdefeature not compiling forwgpu-types. By @KirmesBude in #5149 - Fix the validation of vertex and index ranges. By @nical in #5144 and #5156
- Fix panic when creating a surface while no backend is available. By @wumpf #5166
- Correctly compute minimum buffer size for array-typed
storageanduniformvars. By @jimblandy #5222 - Fix timeout when presenting a surface where no work has been done. By @waywardmonkeys in #5200
- Fix registry leaks with de-duplicated resources. By @nical in #5244
- Fix linking when targeting android. By @ashdnazg in #5326.
- Failing to set the device lost closure will call the closure before returning. By @bradwerth in #5358.
- Fix deadlocks caused by recursive read-write lock acquisitions #5426.
- Remove exposed C symbols (
extern "C"+ [no_mangle]) from RenderPass & ComputePass recording. By @wumpf in #5409. - Fix surfaces being only compatible with first backend enabled on an instance, causing failures when manually specifying an adapter. By @Wumpf in #5535.
- Clean up weak references to texture views and bind groups. By @xiaopengli89 #5595.
- In spv-in, remove unnecessary "gl_PerVertex" name check so unused builtins will always be skipped. Prevents validation errors caused by capability requirements of these builtins #4915. By @Imberflur in #5227.
- In spv-out, check for acceleration and ray-query types when enabling ray-query extension to prevent validation error. By @Vecvec in #5463
- Add a limit for curly brace nesting in WGSL parsing, plus a note about stack size requirements. By @ErichDonGubler in #5447.
- In hlsl-out, fix accesses on zero value expressions by generating helper functions for
Expression::ZeroValue. By @Imberflur in #5587. - Fix behavior of
extractBitsandinsertBitswhenoffset + countoverflows the bit width. By @cwfitzgerald in #5305 - Fix behavior of integer
clampwhenminargument >maxargument. By @cwfitzgerald in #5300. - Fix
TypeInner::scalar_widthto be consistent with the rest of the codebase and return values in bytes not bits. By @atlv24 in #5532.
- GLSL 410 does not support layout(binding = ...), enable only for GLSL 420. By @bes in #5357
- Fixes for being able to use an OpenGL 4.1 core context provided by macOS with wgpu. By @bes in #5331.
- Fix crash when holding multiple devices on wayland/surfaceless. By @ashdnazg in #5351.
- Fix
first_instancegetting ignored in draw indexed whenARB_shader_draw_parametersfeature is present andbase_vertexis 0. By @valaphee in #5482
- Set object labels when the DEBUG flag is set, even if the VALIDATION flag is disabled. By @DJMcNab in #5345.
- Add safety check to
wgpu_hal::vulkan::CommandEncoderto make surediscard_encodingis not called in the closed state. By @villuna in #5557 - Fix SPIR-V type capability requests to not depend on
LocalTypecaching. By @atlv24 in #5590 - Upgrade
ashto0.38. By @MarijnS95 in #5504.
- Fix intermittent crashes on Linux in the
multithreaded_computetest. By @jimblandy in #5129. - Refactor tests to read feature flags by name instead of a hardcoded hexadecimal u64. By @atlv24 in #5155.
- Add test that verifies that we can drop the queue before using the device to create a command encoder. By @Davidster in #5211
- Don't depend on bind group and bind group layout entry order in backends. This caused incorrect severely incorrect command execution and, in some cases, crashes. By @ErichDonGubler in #5421.
- Properly clean up all write_buffer/texture temporary resources. By @robtfm in #5413.
- Fix deadlock in certain situations when mapping buffers using
wgpu-profiler. By @cwfitzgerald in #5517
- Correctly pass through timestamp queries to WebGPU. By @cwfitzgerald in #5527.
This release includes wgpu, wgpu-core, and wgpu-hal. All other crates are unchanged.
--cfg=web_sys_unstable_apis is no longer needed in your RUSTFLAGS to compile for WebGPU!!!
While WebGPU's javascript api is stable in the browsers, the web_sys bindings for WebGPU are still improving. As such they are hidden behind the special cfg --cfg=web_sys_unstable_apis and are not available by default. Everyone who wanted to use our WebGPU backend needed to enable this cfg in their RUSTFLAGS. This was very inconvenient and made it hard to use WebGPU, especially when WebGPU is enabled by default. Additionally, the unstable APIs don't adhere to semver, so there were repeated breakages.
To combat this problem we have decided to vendor the web_sys bindings for WebGPU within the crate. Notably we are not forking the bindings, merely vendoring, so any improvements we make to the bindings will be contributed directly to upstream web_sys.
By @cwfitzgerald in #5325.
- Fix an issue where command encoders weren't properly freed if an error occurred during command encoding. By @ErichDonGubler in #5251.
- Fix incorrect validation causing all indexed draws on render bundles to fail. By @wumpf in #5430.
- Fix linking error when targeting android without
winit. By @ashdnazg in #5326.
This release includes wgpu, wgpu-core, wgpu-hal, wgpu-types, and naga. All other crates are unchanged.
wgpu::Idnow implementsPartialOrd/Ordallowing it to be put inBTreeMaps. By @cwfitzgerald and @9291Sam in #5176
- Log an error when OpenGL texture format heuristics fail. By @PolyMeilex in #5266
- Learned to generate acceleration structure types. By @JMS55 in #5261
- Fix link in
wgpu::Instance::create_surfacedocumentation. By @HexoKnight in #5280. - Fix typo in
wgpu::CommandEncoder::clear_bufferdocumentation. By @PWhiddy in #5281. Surfaceconfiguration incorrectly claimed thatwgpu::Instance::create_surfacewas unsafe. By @hackaugusto in #5265.
- Device lost callbacks are invoked when replaced and when global is dropped. By @bradwerth in #5168
- Fix performance regression when allocating a large amount of resources of the same type. By @nical in #5229
- Fix docs.rs wasm32 builds. By @cwfitzgerald in #5310
- Improve error message when binding count limit hit. By @hackaugusto in #5298
- Remove an unnecessary
cloneduring GLSL shader ingestion. By @a1phyr in #5118. - Fix missing validation for
Device::clear_bufferwhereoffset + size > buffer.sizewas not checked whensizewas omitted. By @ErichDonGubler in #5282.
- Fix
panic!when droppingInstancewithoutInstanceFlags::VALIDATION. By @hakolao in #5134
- Fix internal format for the
Etc2Rgba8Unormformat. By @andristarr in #5178 - Try to load
libX11.so.6in addition tolibX11.soon linux. #5307 - Make use of
GL_EXT_texture_shadow_lodto support sampling a cube depth texture with an explicit LOD. By @cmrschwarz in #5171.
- Fix code generation from nested loops. By @cwfitzgerald and @teoxoy in #5311
This release includes wgpu and wgpu-hal. The rest of the crates are unchanged since 0.19.0.
- Properly register all swapchain buffers to prevent error on surface present. By @dtzxporter in #5091
- Check for extra null states when creating resources. By @nical in #5096
- Fix depth-only and stencil-only views causing crashes. By @teoxoy in #5100
- In Surface::configure and Surface::present on Windows, fix the current GL context not being unset when releasing the lock that guards access to making the context current. This was causing other threads to panic when trying to make the context current. By @Imberflur in #5087.
- Improve error message when compiling WebGPU backend on wasm without the
web_sys_unstable_apisset. By @rukai in #5104
- Document Wayland specific behavior related to
SurfaceTexture::present. By @i509VCB in #5093.
This release includes:
wgpuwgpu-corewgpu-halwgpu-typeswgpu-infonaga(skipped from 0.14 to 0.19)naga-cli(skipped from 0.14 to 0.19)d3d12(skipped from 0.7 to 0.19)
Large refactoring of wgpu’s internals aiming at reducing lock contention, and providing better performance when using wgpu on multiple threads.
By @gents83 in #3626 and thanks also to @jimblandy, @nical, @Wumpf, @Elabajaba & @cwfitzgerald
All of wgpu's public dependencies are now re-exported at the top level so that users don't need to take their own dependencies. This includes:
- wgpu-core
- wgpu-hal
- naga
- raw_window_handle
- web_sys
Enabling webgl no longer removes the webgpu backend.
Instead, there's a new (default enabled) webgpu feature that allows to explicitly opt-out of webgpu if so desired.
If both webgl & webgpu are enabled, wgpu::Instance decides upon creation whether to target wgpu-core/WebGL or WebGPU.
This means that adapter selection is not handled as with regular adapters, but still allows to decide at runtime whether
webgpu or the webgl backend should be used using a single wasm binary.
By @wumpf in #5044
The naga-ir feature has been added to allow you to add naga module shaders without guessing about what other features needed to be enabled to get access to it.
By @cwfitzgerald in #5063.
This feature allowed you to call global_id on any wgpu opaque handle to get a unique hashable identity for the given resource. This is now available without the feature flag.
By @cwfitzgerald in #4841.
wgpu now exposes backend feature for the Direct3D 12 (dx12) and Metal (metal) backend. These are enabled by default, but don't do anything when not targeting the corresponding OS.
By @daxpedda in #4815.
This backend had no functionality, and with the recent support for GL on Desktop, which allows wgpu to run on older devices, there was no need to keep this backend. By @valaphee in #4828.
This adds a way to allow a Vulkan driver which is non-compliant per VK_KHR_driver_properties to be enumerated. This is intended for testing new Vulkan drivers which are not Vulkan compliant yet.
By @i509VCB in #4754.
Previously, DeviceExt::create_texture_with_data only allowed data to be provided in layer major order. There is now a order parameter which allows you to specify if the data is in layer major or mip major order.
let tex = ctx.device.create_texture_with_data(
&queue,
&descriptor,
+ wgpu::util::TextureDataOrder::LayerMajor,
src_data,
);By @cwfitzgerald in #4780.
It is now possible to safely create a wgpu::Surface with wgpu::Instance::create_surface() by letting wgpu::Surface hold a lifetime to window.
Passing an owned value window to Surface will return a wgpu::Surface<'static>.
All possible safe variants (owned windows and web canvases) are grouped using wgpu::SurfaceTarget.
Conversion to wgpu::SurfaceTarget is automatic for any type implementing raw-window-handle's HasWindowHandle & HasDisplayHandle traits, i.e. most window types.
For web canvas types this has to be done explicitly:
let surface: wgpu::Surface<'static> = instance.create_surface(wgpu::SurfaceTarget::Canvas(my_canvas))?;All unsafe variants are now grouped under wgpu::Instance::create_surface_unsafe which takes the
wgpu::SurfaceTargetUnsafe enum and always returns wgpu::Surface<'static>.
In order to create a wgpu::Surface<'static> without passing ownership of the window use
wgpu::SurfaceTargetUnsafe::from_window:
let surface = unsafe {
instance.create_surface_unsafe(wgpu::SurfaceTargetUnsafe::from_window(&my_window))?
};The easiest way to make this code safe is to use shared ownership:
let window: Arc<winit::Window>;
// ...
let surface = instance.create_surface(my_window.clone())?;All platform specific surface creation using points have moved into SurfaceTargetUnsafe as well.
For example:
Safety by @daxpedda in #4597 Unification by @wumpf in #4984
Abstract types make numeric literals easier to use, by automatically converting literals and other constant expressions from abstract numeric types to concrete types when safe and necessary. For example, to build a vector of floating-point numbers, Naga previously made you write:
vec3<f32>(1.0, 2.0, 3.0)With this change, you can now simply write:
vec3<f32>(1, 2, 3)Even though the literals are abstract integers, Naga recognizes
that it is safe and necessary to convert them to f32 values in
order to build the vector. You can also use abstract values as
initializers for global constants and global and local variables,
like this:
var unit_x: vec2<f32> = vec2(1, 0);The literals 1 and 0 are abstract integers, and the expression
vec2(1, 0) is an abstract vector. However, Naga recognizes that
it can convert that to the concrete type vec2<f32> to satisfy
the given type of unit_x.
The WGSL specification permits abstract integers and
floating-point values in almost all contexts, but Naga's support
for this is still incomplete. Many WGSL operators and builtin
functions are specified to produce abstract results when applied
to abstract inputs, but for now Naga simply concretizes them all
before applying the operation. We will expand Naga's abstract type
support in subsequent pull requests.
As part of this work, the public types naga::ScalarKind and
naga::Literal now have new variants, AbstractInt and AbstractFloat.
By @jimblandy in #4743, #4755.
This allows us to support WebGPU and WebGL in the same binary.
- let adapters: Vec<Adapter> = instance.enumerate_adapters(wgpu::Backends::all()).collect();
+ let adapters: Vec<Adapter> = instance.enumerate_adapters(wgpu::Backends::all());By @wumpf in #5044
This is a forward looking change, as we plan to add more information to the MaintainResult in the future.
This enum has the same data as the boolean, but with some useful helper functions.
- let queue_finished: bool = device.poll(wgpu::Maintain::Wait);
+ let queue_finished: bool = device.poll(wgpu::Maintain::Wait).is_queue_empty();By @cwfitzgerald in #5053
- Added
DownlevelFlags::VERTEX_AND_INSTANCE_INDEX_RESPECTS_RESPECTIVE_FIRST_VALUE_IN_INDIRECT_DRAWto know if@builtin(vertex_index)and@builtin(instance_index)will respect thefirst_vertex/first_instancein indirect calls. If this is not present, both will always start counting from 0. Currently enabled on all backends except DX12. By @cwfitzgerald in #4722. - Added support for the
FLOAT32_FILTERABLEfeature (web and native, corresponds to WebGPU'sfloat32-filterable). By @almarklein in #4759. - GPU buffer memory is released during "lose the device". By @bradwerth in #4851.
- wgpu and wgpu-core cargo feature flags are now documented on docs.rs. By @wumpf in #4886.
- DeviceLostClosure is guaranteed to be invoked exactly once. By @bradwerth in #4862.
- Log vulkan validation layer messages during instance creation and destruction: By @exrook in #4586.
TextureFormat::block_sizeis deprecated, useTextureFormat::block_copy_sizeinstead: By @wumpf in #4647.- Rename of
DispatchIndirect,DrawIndexedIndirect, andDrawIndirecttypes in thewgpu::utilmodule toDispatchIndirectArgs,DrawIndexedIndirectArgs, andDrawIndirectArgs. By @cwfitzgerald in #4723. - Make the size parameter of
encoder.clear_bufferanOption<u64>instead ofOption<NonZero<u64>>. By @nical in #4737. - Reduce the
infolog level noise. By @nical in #4769, #4711 and #4772 - Rename
features&limitsfields ofDeviceDescriptortorequired_features&required_limits. By @teoxoy in #4803. SurfaceConfigurationnow exposesdesired_maximum_frame_latencywhich was previously hard-coded to 2. By setting it to 1 you can reduce latency under the risk of making GPU & CPU work sequential. Currently, on DX12 this affects theMaximumFrameLatency, on all other backends except OpenGL the size of the swapchain (on OpenGL this has no effect). By @emilk & @wumpf in #4899
@builtin(instance_index)now properly reflects the range provided in the draw call instead of always counting from 0. By @cwfitzgerald in #4722.- Desktop GL now supports
POLYGON_MODE_LINEandPOLYGON_MODE_POINT. By @valaphee in #4836.
- Naga's WGSL front end now allows operators to produce values with abstract types, rather than concretizing thir operands. By @jimblandy in #4850 and #4870.
- Naga's WGSL front and back ends now have experimental support for 64-bit floating-point literals:
1.0lfdenotes anf64value. There has been experimental support for anf64type for a while, but until now there was no syntax for writing literals with that type. As before, Naga module validation rejectsf64values unlessnaga::valid::Capabilities::FLOAT64is requested. By @jimblandy in #4747. - Naga constant evaluation can now process binary operators whose operands are both vectors. By @jimblandy in #4861.
- Add
--bulk-validateoption to Naga CLI. By @jimblandy in #4871. - Naga's
cargo xtask validatenow runs validation jobs in parallel, using the jobserver protocol to limit concurrency, and offers avalidate allsubcommand, which runs all available validation types. By @jimblandy in #4902. - Remove
spanandvalidatefeatures. Always fully validate shader modules, and always track source positions for use in error messages. By @teoxoy in #4706. - Introduce a new
Scalarstruct type for use in Naga's IR, and update all frontend, middle, and backend code appropriately. By @jimblandy in #4673. - Add more metal keywords. By @fornwall in #4707.
- Add a new
naga::Literalvariant,I64, for signed 64-bit literals. #4711. - Emit and init
structmember padding always. By @ErichDonGubler in #4701. - In WGSL output, always include the
isuffix oni32literals. By @jimblandy in #4863. - In WGSL output, always include the
fsuffix onf32literals. By @jimblandy in #4869.
BufferMappedRangetrait is nowWasmNotSendSync, i.e. it isSend/Syncif not on wasm orfragile-send-sync-non-atomic-wasmis enabled. By @wumpf in #4818.- Align
wgpu_types::CompositeAlphaModeserde serialization to spec. By @littledivy in #4940. - Fix error message of
ConfigureSurfaceError::TooLarge. By @Dinnerbone in #4960. - Fix dropping of
DeviceLostCallbackCparams. By @bradwerth in #5032. - Fixed a number of panics. By @nical in #4999, #5014, #5024, #5025, #5026, #5027, #5028 and #5042.
- No longer validate surfaces against their allowed extent range on configure. This caused warnings that were almost impossible to avoid. As before, the resulting behavior depends on the compositor. By @wumpf in #4796.
- Fixed D3D12_SUBRESOURCE_FOOTPRINT calculation for block compressed textures which caused a crash with
Queue::write_textureon DX12. By @DTZxPorter in #4990.
- Use
VK_EXT_robustness2only when not using an outdated intel iGPU driver. By @TheoDulka in #4602.
- Allow calling
BufferSlice::get_mapped_rangemultiple times on the same buffer slice (instead of throwing a Javascript exception). By @DouglasDwyer in #4726.
- Create a hidden window per
wgpu::Instanceinstead of sharing a global one. By @Zoxc in #4603
- Make module compaction preserve the module's named types, even if they are unused. By @jimblandy in #4734.
- Improve algorithm used by module compaction. By @jimblandy in #4662.
- When reading GLSL, fix the argument types of the double-precision floating-point overloads of the
dot,reflect,distance, andldexpbuiltin functions. Correct the WGSL generated for constructing 64-bit floating-point matrices. Add tests for all the above. By @jimblandy in #4684. - Allow Naga's IR types to represent matrices with elements elements of any scalar kind. This makes it possible for Naga IR types to represent WGSL abstract matrices. By @jimblandy in #4735.
- Preserve the source spans for constants and expressions correctly across module compaction. By @jimblandy in #4696.
- Record the names of WGSL
aliasdeclarations in Naga IRTypes. By @jimblandy in #4733.
- Allow the
COPY_SRCusage flag in surface configuration. By @Toqozz in #4852.
- remove winit dependency from hello-compute example. By @psvri in #4699
- hello-compute example fix failure with
wgpu error: Validation Errorif arguments are missing. By @vilcans in #4939. - Made the examples page not crash on Chrome on Android, and responsive to screen sizes. By @Dinnerbone in #4958.
This release includes naga version 0.14.2. The crates wgpu-core, wgpu-hal are still at 0.18.1 and the crates wgpu and wgpu-types are still at 0.18.0.
- When evaluating const-expressions and generating SPIR-V, properly handle
Composeexpressions whose operands areSplatexpressions. Such expressions are created and marked as constant by the constant evaluator. By @jimblandy in #4695.
(naga version 0.14.1)
- Fix panic in
Surface::configurein debug builds. By @cwfitzgerald in #4635 - Fix crash when all the following are true: By @teoxoy in ##4642
- Passing a naga module directly to
Device::create_shader_module. InstanceFlags::DEBUGis enabled.
- Passing a naga module directly to
- Always use HLSL 2018 when using DXC to compile HLSL shaders. By @daxpedda in #4629
- In Metal Shading Language output, fix issue where local variables were sometimes using variable names from previous functions. By @DJMcNab in #4594
For naga changelogs at or before v0.14.0. See naga's changelog.
We now support OpenGL on Windows! This brings support for a vast majority of the hardware that used to be covered by our DX11 backend. As of this writing we support OpenGL 3.3+, though there are efforts to reduce that further.
This allows us to cover the last 12 years of Intel GPUs (starting with Ivy Bridge; aka 3xxx), and the last 16 years of AMD (starting with Terascale; aka HD 2000) / NVidia GPUs (starting with Tesla; aka GeForce 8xxx).
By @Zoxc in #4248
Timestamp queries are now supported on both Metal and Desktop OpenGL. On Apple chips on Metal, they only support timestamp queries in command buffers or in the renderpass descriptor, they do not support them inside a pass.
Metal: By @Wumpf in #4008 OpenGL: By @Zoxc in #4267
Addition of the TimestampWrites type to compute and render pass descriptors to allow profiling on tilers which do not support timestamps inside passes.
Added an example to demonstrate the various kinds of timestamps.
Additionally, metal now supports timestamp queries!
By @FL33TW00D & @wumpf in #3636.
We now support binary occlusion queries! This allows you to determine if any of the draw calls within the query drew any pixels.
Use the new occlusion_query_set field on RenderPassDescriptor to give a query set that occlusion queries will write to.
let mut rpass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
// ...
+ occlusion_query_set: Some(&my_occlusion_query_set),
});Within the renderpass do the following to write the occlusion query results to the query set at the given index:
rpass.begin_occlusion_query(index);
rpass.draw(...);
rpass.draw(...);
rpass.end_occlusion_query();These are binary occlusion queries, so the result will be either 0 or an unspecified non-zero value.
By @Valaphee in #3402
// WGSL constant expressions are now supported!
const BLAH: u32 = 1u + 1u;
// `rgb10a2uint` and `bgra8unorm` can now be used as a storage image format.
var image: texture_storage_2d<rgb10a2uint, write>;
var image: texture_storage_2d<bgra8unorm, write>;
// You can now use dual source blending!
struct FragmentOutput{
@location(0) source1: vec4<f32>,
@location(0) @second_blend_source source2: vec4<f32>,
}
// `modf`/`frexp` now return structures
let result = modf(1.5);
result.fract == 0.5;
result.whole == 1.0;
let result = frexp(1.5);
result.fract == 0.75;
result.exponent == 2i;
// `modf`/`frexp` are currently disabled on GLSL and SPIR-V input.// Cannot get pointer to a workgroup variable
fn func(p: ptr<workgroup, u32>); // ERROR
// Cannot create Inf/NaN through constant expressions
const INF: f32 = 3.40282347e+38 + 1.0; // ERROR
const NAN: f32 = 0.0 / 0.0; // ERROR
// `outerProduct` function removed
// Error on repeated or missing `@workgroup_size()`
@workgroup_size(1) @workgroup_size(2) // ERROR
fn compute_main() {}
// Error on repeated attributes.
fn fragment_main(@location(0) @location(0) location_0: f32) // ERRORwgpu::Operations::store used to be an underdocumented boolean value,
causing misunderstandings of the effect of setting it to false.
The API now more closely resembles WebGPU which distinguishes between store and discard,
see WebGPU spec on GPUStoreOp.
// ...
depth_ops: Some(wgpu::Operations {
load: wgpu::LoadOp::Clear(1.0),
- store: false,
+ store: wgpu::StoreOp::Discard,
}),
// ...By @wumpf in #4147
The instance descriptor grew two more fields: flags and gles_minor_version.
flags allow you to toggle the underlying api validation layers, debug information about shaders and objects in capture programs, and the ability to discard labels
gles_minor_version is a rather niche feature that allows you to force the GLES backend to use a specific minor version, this is useful to get ANGLE to enable more than GLES 3.0.
let instance = wgpu::Instance::new(InstanceDescriptor {
...
+ flags: wgpu::InstanceFlags::default()
+ gles_minor_version: wgpu::Gles3MinorVersion::Automatic,
});gles_minor_version: By @PJB3005 in #3998
flags: By @nical in #4230
- Added the following examples: By @JustAnotherCodemonkey in #3885.
Our testing harness was completely revamped and now automatically runs against all gpus in the system, shows the expected status of every test, and is tolerant to flakes.
Additionally, we have filled out our CI to now run the latest versions of WARP and Mesa. This means we can test even more features on CI than before.
By @cwfitzgerald in #3873
The angle feature flag has to be set for the GLES backend to be enabled on Windows & macOS.
By @teoxoy in #4185
- Omit texture store bound checks since they are no-ops if out of bounds on all APIs. By @teoxoy in #3975
- Validate
DownlevelFlags::READ_ONLY_DEPTH_STENCIL. By @teoxoy in #4031 - Add validation in accordance with WebGPU
setViewportvalid usage forx,yandthis.[[attachment_size]]. By @James2022-rgb in #4058 wgpu::CreateSurfaceErrorandwgpu::RequestDeviceErrornow give details of the failure, but no longer implementPartialEqand cannot be constructed. By @kpreid in #4066 and #4145- Make
WGPU_POWER_PREF=nonea valid value. By @fornwall in 4076 - Support dual source blending in OpenGL ES, Metal, Vulkan & DX12. By @freqmod in 4022
- Add stub support for device destroy and device validity. By @bradwerth in 4163 and in 4212
- Add trace-level logging for most entry points in wgpu-core By @nical in 4183
- Add
Rgb10a2Uintformat. By @teoxoy in 4199 - Validate that resources are used on the right device. By @nical in 4207
- Expose instance flags.
- Add support for the bgra8unorm-storage feature. By @jinleili and @nical in #4228
- Calls to lost devices now return
DeviceError::Lostinstead ofDeviceError::Invalid. By @bradwerth in #4238 - Let the
"strict_asserts"feature enable check that wgpu-core's lock-ordering tokens are unique per thread. By @jimblandy in #4258 - Allow filtering labels out before they are passed to GPU drivers by @nical in https://github.com/gfx-rs/wgpu/pull/4246
DeviceLostClosurecallback mechanism provided so user agents can resolveGPUDevice.lostPromises at the appropriate time by @bradwerth in #4645
- Rename
wgpu_hal::vulkan::Instance::required_extensionstodesired_extensions. By @jimblandy in #4115 - Don't bother calling
vkFreeCommandBufferswhenvkDestroyCommandPoolwill take care of that for us. By @jimblandy in #4059
- Bump
gpu-allocatorto 0.23. By @Elabajaba in #4198
- Use WGSL for VertexFormat example types. By @ScanMountGoat in #4035
- Fix description of
Features::TEXTURE_COMPRESSION_ASTC_HDRin #4157
- Derive storage bindings via
naga::StorageAccessinstead ofnaga::GlobalUse. By @teoxoy in #3985. Queue::on_submitted_work_donecallbacks will now always be called after all previousBufferSlice::map_asynccallbacks, even when there are no active submissions. By @cwfitzgerald in #4036.- Fix
cleartexture views being leaked whenwgpu::SurfaceTextureis dropped before it is presented. By @rajveermalviya in #4057. - Add
Feature::SHADER_UNUSED_VERTEX_OUTPUTto allow unused vertex shader outputs. By @Aaron1011 in #4116. - Fix a panic in
surface_configure. By @nical in #4220 and #4227 - Pipelines register their implicit layouts in error cases. By @bradwerth in #4624
- Better handle explicit destruction of textures and buffers. By @nical in #4657
- Fix enabling
wgpu::Features::PARTIALLY_BOUND_BINDING_ARRAYnot being actually enabled in vulkan backend. By @39ali in#3772. - Don't pass
vk::InstanceCreateFlags::ENUMERATE_PORTABILITY_KHRunless theVK_KHR_portability_enumerationextension is available. By @jimblandy in#4038. - Enhancement of [#4038], using ash's definition instead of hard-coded c_str. By @hybcloud in#4044.
- Enable vulkan presentation on (Linux) Intel Mesa >= v21.2. By @flukejones in#4110
- DX12 doesn't support `Features::POLYGON_MODE_POINT``. By @teoxoy in #4032.
- Set
Features::VERTEX_WRITABLE_STORAGEbased on the right feature level. By @teoxoy in #4033.
- Ensure that MTLCommandEncoder calls endEncoding before it is deallocated. By @bradwerth in #4023
- Ensure that limit requests and reporting is done correctly. By @OptimisticPeach in #4107
- Validate usage of polygon mode. By @teoxoy in #4196
- enable/disable blending per attachment only when available (on ES 3.2 or higher). By @teoxoy in #4234
- Add an overview of
RenderPassand how render state works. By @kpreid in #4055
- Created
wgpu-example::utilsmodule to contain misc functions and such that are common code but aren't part of the example framework. Add to it the functionsoutput_image_wasmandoutput_image_native, both for outputtingVec<u8>RGBA images either to the disc or the web page. By @JustAnotherCodemonkey in #3885. - Removed
captureexample as it had issues (did not run on wasm) and has been replaced byrender-to-texture(see above). By @JustAnotherCodemonkey in #3885.
- Fix x11 hang while resizing on vulkan. @Azorlogh in #4184.
- Add
get_mapped_range_as_array_bufferfor faster buffer read-backs in wasm builds. By @ryankaplan in [#4042] (#4042).
- Fix panic on resize when using DX12. By @cwfitzgerald in #4106
- Suppress validation error caused by OBS layer. This was also fixed upstream. By @cwfitzgerald in #4002
- Work around bug in nvidia's vkCmdFillBuffer implementation. By @cwfitzgerald in #4132.
This is the first release that featured wgpu-info as a binary crate for getting information about what devices wgpu sees in your system. It can dump the information in both human readable format and json.
This release was fairly minor as breaking changes go.
Up until this point, wgpu has made the assumption that threads do not exist on wasm. With the rise of libraries like wasm_thread making it easier and easier to do wasm multithreading this assumption is no longer sound. As all wgpu objects contain references into the JS heap, they cannot leave the thread they started on.
As we understand that this change might be very inconvenient for users who don't care about wasm threading, there is a crate feature which re-enables the old behavior: fragile-send-sync-non-atomic-wasm. So long as you don't compile your code with -Ctarget-feature=+atomics, Send and Sync will be implemented again on wgpu types on wasm. As the name implies, especially for libraries, this is very fragile, as you don't know if a user will want to compile with atomics (and therefore threads) or not.
By @daxpedda in #3691
The power_preference field of RequestAdapterOptions is now optional. If it is PowerPreference::None, we will choose the first available adapter, preferring GPU adapters over CPU adapters.
By @Aaron1011 in #3903
Removed the backend_bits parameter from initialize_adapter_from_env and initialize_adapter_from_env_or_default. If you want to limit the backends used by this function, only enable the wanted backends in the instance.
Added a compatible surface parameter, to ensure the given device is able to be presented onto the given surface.
- wgpu::util::initialize_adapter_from_env(instance, backend_bits);
+ wgpu::util::initialize_adapter_from_env(instance, Some(&compatible_surface));By @fornwall in #3904 and #3905
- Change
AdapterInfo::{device,vendor}to beu32instead ofusize. By @ameknite in #3760
- Added support for importing external buffers using
buffer_from_raw(Dx12, Metal, Vulkan) andcreate_buffer_from_hal. By @AdrianEddy in #3355
- Work around Vulkan-ValidationLayers#5671 by ignoring reports of violations of VUID-vkCmdEndDebugUtilsLabelEXT-commandBuffer-01912. By @jimblandy in #3809.
- Empty scissor rects are allowed now, matching the specification. by @PJB3005 in #3863.
- Add back components info to
TextureFormats. By @teoxoy in #3843. - Add
get_mapped_range_as_array_bufferfor faster buffer read-backs in wasm builds. By @ryankaplan in [#4042] (#4042).
- Better documentation for draw, draw_indexed, set_viewport and set_scissor_rect. By @genusistimelord in #3860
- Fix link to
GPUVertexBufferLayout. By @fornwall in #3906 - Document feature requirements for
DEPTH32FLOAT_STENCIL8by @ErichDonGubler in #3734. - Flesh out docs. for
AdapterInfo::{device,vendor}by @ErichDonGubler in #3763. - Spell out which sizes are in bytes. By @jimblandy in #3773.
- Validate that
descriptor.usageis not empty increate_bufferby @nical in #3928 - Update
max_bindings_per_bind_grouplimit to reflect spec changes by @ErichDonGubler and @nical in #3943 #3942 - Add better docs for
Limits, listing the actual limits returned bydownlevel_defaultsanddownlevel_webgl2_defaultsby @JustAnotherCodemonkey in #3988
- Fix order of arguments to glPolygonOffset by @komadori in #3783.
- Fix OpenGL/EGL backend not respecting non-sRGB texture formats in
SurfaceConfiguration. by @liquidev in #3817 - Make write- and read-only marked buffers match non-readonly layouts. by @fornwall in #3893
- Fix leaking X11 connections. by @wez in #3924
- Fix ASTC feature selection in the webgl backend. by @expenses in #3934
- Fix Multiview to disable validation of TextureViewDimension and ArrayLayerCount. By @MalekiRe in #3779.
- Fix incorrect aspect in barriers when using emulated Stencil8 textures. By @cwfitzgerald in #3833.
- Implement depth-clip-control using depthClamp instead of VK_EXT_depth_clip_enable. By @AlbinBernhardssonARM #3892.
- Fix enabling
wgpu::Features::PARTIALLY_BOUND_BINDING_ARRAYnot being actually enabled in vulkan backend. By @39ali in#3772.
- Fix renderpasses being used inside of renderpasses. By @cwfitzgerald in #3828
- Support (simulated) visionOS. By @jinleili in #3883
- Disable suballocation on Intel Iris(R) Xe. By @xiaopengli89 in #3668
- Change the
max_buffer_sizelimit fromu64::MAXtoi32::MAX. By @nical in #4020
- Use
get_preferred_canvas_format()to fillformatsofSurfaceCapabilities. By @jinleili in #3744
- Publish examples to wgpu.rs on updates to trunk branch instead of gecko. By @paul-hansen in #3750
- Ignore the exception values generated by the winit resize event. By @jinleili in #3916
- Make the
Idtype that is exposed when using theexpose-idsfeature implementSendandSyncagain. This was unintentionally changed by the v0.16.0 release and is now fixed.
- Increase the
max_storage_buffers_per_shader_stageandmax_storage_textures_per_shader_stagelimits based on what the hardware supports. by @Elabajaba in [#3798]#3798
- Fix missing 4X MSAA support on some OpenGL backends. By @emilk in #3780
- Fix crash on dropping
wgpu::CommandBuffer. By @wumpf in #3726. - Use
u32s internally for bind group indices, rather thanu8. By @ErichDonGubler in #3743.
- Fix crash when calling
create_surface_from_canvas. By @grovesNL in #3718
type has been replaced with alias to match with upstream WebGPU.
- type MyType = vec4<u32>;
+ alias MyType = vec4<u32>;The TextureFormat::describe function was removed in favor of separate functions: block_dimensions, is_compressed, is_srgb, required_features, guaranteed_format_features, sample_type and block_size.
- let block_dimensions = format.describe().block_dimensions;
+ let block_dimensions = format.block_dimensions();
- let is_compressed = format.describe().is_compressed();
+ let is_compressed = format.is_compressed();
- let is_srgb = format.describe().srgb;
+ let is_srgb = format.is_srgb();
- let required_features = format.describe().required_features;
+ let required_features = format.required_features();Additionally guaranteed_format_features now takes a set of features to assume are enabled.
- let guaranteed_format_features = format.describe().guaranteed_format_features;
+ let guaranteed_format_features = format.guaranteed_format_features(device.features());Additionally sample_type and block_size now take an optional TextureAspect and return Options.
- let sample_type = format.describe().sample_type;
+ let sample_type = format.sample_type(None).expect("combined depth-stencil format requires specifying a TextureAspect");
- let block_size = format.describe().block_size;
+ let block_size = format.block_size(None).expect("combined depth-stencil format requires specifying a TextureAspect");By @teoxoy in #3436
Buffers used as the destination argument of CommandEncoder::resolve_query_set now have to contain the QUERY_RESOLVE usage instead of the COPY_DST usage.
let destination = device.create_buffer(&wgpu::BufferDescriptor {
// ...
- usage: wgpu::BufferUsages::COPY_DST | wgpu::BufferUsages::MAP_READ,
+ usage: wgpu::BufferUsages::QUERY_RESOLVE | wgpu::BufferUsages::MAP_READ,
mapped_at_creation: false,
});
command_encoder.resolve_query_set(&query_set, query_range, &destination, destination_offset);By @JolifantoBambla in #3489
The following Features have been renamed.
SHADER_FLOAT16->SHADER_F16SHADER_FLOAT64->SHADER_F64SHADER_INT16->SHADER_I16TEXTURE_COMPRESSION_ASTC_LDR->TEXTURE_COMPRESSION_ASTCWRITE_TIMESTAMP_INSIDE_PASSES->TIMESTAMP_QUERY_INSIDE_PASSES
By @teoxoy in #3534
Anisotropic filtering has been brought in line with the spec. The anisotropic clamp is now a u16 (was a Option<u8>) which must be at least 1.
If the anisotropy clamp is not 1, all the filters in a sampler must be Linear.
SamplerDescriptor {
- anisotropic_clamp: None,
+ anisotropic_clamp: 1,
}By @cwfitzgerald in #3610.
Some texture format names have changed to get back in line with the spec.
- TextureFormat::Bc6hRgbSfloat
+ TextureFormat::Bc6hRgbFloatBy @cwfitzgerald in #3671.
- Change type of
mip_level_countandarray_layer_count(members ofTextureViewDescriptorandImageSubresourceRange) fromOption<NonZeroU32>toOption<u32>. By @teoxoy in #3445 - Change type of
bytes_per_rowandrows_per_image(members ofImageDataLayout) fromOption<NonZeroU32>toOption<u32>. By @teoxoy in #3529 - On Web,
Instance::create_surface_from_canvas()andcreate_surface_from_offscreen_canvas()now take the canvas by value. By @daxpedda in #3690
- Added feature flags for ray-tracing (currently only hal):
RAY_QUERYandRAY_TRACING@daniel-keitel (started by @expenses) in #3507
- Implemented basic ray-tracing api for acceleration structures, and ray-queries @daniel-keitel (started by @expenses) in #3507
- Added basic ray-tracing api for acceleration structures, and ray-queries @daniel-keitel (started by @expenses) in #3507
- Added
TextureFormatFeatureFlags::MULTISAMPLE_X16. By @Dinnerbone in #3454 - Added
BufferUsages::QUERY_RESOLVE. By @JolifantoBambla in #3489 - Support stencil-only views and copying to/from combined depth-stencil textures. By @teoxoy in #3436
- Added
Features::SHADER_EARLY_DEPTH_TEST. By @teoxoy in #3494 - All
fxhashdependencies have been replaced withrustc-hash. By @james7132 in #3502 - Allow copying of textures with copy-compatible formats. By @teoxoy in #3528
- Improve attachment related errors. By @cwfitzgerald in #3549
- Make error descriptions all upper case. By @cwfitzgerald in #3549
- Don't include ANSI terminal color escape sequences in shader module validation error messages. By @jimblandy in #3591
- Report error messages from DXC compile. By @Davidster in #3632
- Error in native when using a filterable
TextureSampleType::Floaton a multisampleBindingType::Texture. By @mockersf in #3686 - On Web, the size of the canvas is adjusted when using
Surface::configure(). If the canvas was given an explicit size (via CSS), this will not affect the visual size of the canvas. By @daxpedda in #3690 - Added
Global::create_render_bundle_error. By @jimblandy in #3746
- Implement the new checks for readonly stencils. By @JCapucho in #3443
- Reimplement
adapter|device_features. By @jinleili in #3428 - Implement
command_encoder_resolve_query_set. By @JolifantoBambla in #3489 - Add support for
Features::RG11B10UFLOAT_RENDERABLE. By @mockersf in #3689
- Set
max_memory_allocation_sizeviaPhysicalDeviceMaintenance3Properties. By @jinleili in #3567 - Silence false-positive validation error about surface resizing. By @seabassjh in #3627
copyTextureToTexturesrc/dst aspects must both refer to all aspects of src/dst format. By @teoxoy in #3431- Validate before extracting texture selectors. By @teoxoy in #3487
- Fix fatal errors (those which panic even if an error handler is set) not including all of the details. By @kpreid in #3563
- Validate shader location clashes. By @emilk in #3613
- Fix surfaces not being dropped until exit. By @benjaminschaaf in #3647
- Fix handling of
Nonevalues fordepth_opsandstencil_opsinRenderPassDescriptor::depth_stencil_attachment. By @niklaskorz in #3660 - Avoid using
WasmAbifunctions for WebGPU backend. By @grovesNL in #3657
- Use typeless formats for textures that might be viewed as srgb or non-srgb. By @teoxoy in #3555
- Set FORCE_POINT_SIZE if it is vertex shader with mesh consist of point list. By @REASY in 3440
- Remove unwraps inside
surface.configure. By @cwfitzgerald in #3585 - Fix
copy_external_image_to_texture,copy_texture_to_textureandcopy_buffer_to_texturenot taking the specified index into account if the target texture is a cube map, 2D texture array or cube map array. By @daxpedda #3641 - Fix disabling of vertex attributes with non-consecutive locations. By @Azorlogh in #3706
- Fix metal erroring on an
array_strideof 0. By @teoxoy in #3538 create_texturereturns an error ifnew_texturereturns NULL. By @jinleili in #3554- Fix shader bounds checking being ignored. By @FL33TW00D in #3603
- Treat
VK_SUBOPTIMAL_KHRasVK_SUCCESSon Android due to rotation issues. By @James2022-rgb in #3525
- Use
BufferUsages::QUERY_RESOLVEinstead ofBufferUsages::COPY_DSTfor buffers used inCommandEncoder::resolve_query_setcalls inmipmapexample. By @JolifantoBambla in #3489
- Fix incorrect mipmap being sampled when using
MinLod <= 0.0andMaxLod >= 32.0or when the fragment shader samples different Lods in the same quad. By @cwfitzgerald in #3610.
- Fix
Vertex buffer is not big enough for the draw call.for ANGLE/Web when rendering with instance attributes on a single instance. By @wumpf in #3596 - Reset all queue state between command buffers in a submit. By @jleibs #3589
- Reset the state of
SAMPLE_ALPHA_TO_COVERAGEon queue reset. By @jleibs #3589
- Fix definition of
NSOperatingSystemVersionto avoid potential crashes. By @grovesNL in #3557
- Enable
WEBGL_debug_renderer_infobefore querying unmasked vendor/renderer to avoid crashing on emscripten in #3519
- Fix for some minor issues in comments on some features. By @Wumpf in #3455
- Improve format MSAA capabilities detection. By @jinleili in #3429
- Update gpu allocator to 0.22. By @Elabajaba in #3447
- Implement
CommandEncoder::clear_buffer. By @raphlinus in #3426
- Re-sort supported surface formats based on srgb-ness. By @cwfitzgerald in #3444
- Fix surface view formats validation error. By @jinleili in #3432
- Fix DXC validation issues when using a custom
dxil_path. By @Elabajaba in #3434
- Unbind vertex buffers at end of renderpass. By @cwfitzgerald in #3459
- Reimplement
{adapter|device}_features. By @jinleili in #3428
- Build for Wasm on docs.rs. By @daxpedda in #3462
All top level constants are now declared with const, catching up with the wgsl spec.
let is no longer allowed at the global scope, only within functions.
-let SOME_CONSTANT = 12.0;
+const SOME_CONSTANT = 12.0;See https://github.com/gfx-rs/naga/blob/master/CHANGELOG.md#v011-2023-01-25 for smaller shader improvements.
The various surface capability functions were combined into a single call that gives you all the capabilities.
- let formats = surface.get_supported_formats(&adapter);
- let present_modes = surface.get_supported_present_modes(&adapter);
- let alpha_modes = surface.get_supported_alpha_modes(&adapter);
+ let caps = surface.get_capabilities(&adapter);
+ let formats = caps.formats;
+ let present_modes = caps.present_modes;
+ let alpha_modes = caps.alpha_modes;Additionally Surface::get_default_config now returns an Option and returns None if the surface isn't supported by the adapter.
- let config = surface.get_default_config(&adapter);
+ let config = surface.get_default_config(&adapter).expect("Surface unsupported by adapter");Instance::create_surface() now returns Result<Surface, CreateSurfaceError> instead of Surface. This allows an error to be returned if the given window is a HTML canvas and obtaining a WebGPU or WebGL 2 context fails. (No other platforms currently report any errors through this path.) By @kpreid in #3052
A new api, Queue::copy_external_image_to_texture, allows you to create wgpu textures from various web image primitives. Specifically from HtmlVideoElement, HtmlCanvasElement, OffscreenCanvas, and ImageBitmap. This provides multiple low-copy ways of interacting with the browser. WebGL is also supported, though WebGL has some additional restrictions, represented by the UNRESTRICTED_EXTERNAL_IMAGE_COPIES downlevel flag. By @cwfitzgerald in #3288
Instance::new() and hub::Global::new() now take an InstanceDescriptor struct which contains both the existing Backends selection as well as a new Dx12Compiler field for selecting which Dx12 shader compiler to use.
- let instance = Instance::new(wgpu::Backends::all());
+ let instance = Instance::new(wgpu::InstanceDescriptor {
+ backends: wgpu::Backends::all(),
+ dx12_shader_compiler: wgpu::Dx12Compiler::Fxc,
+ });Instance now also also implements Default, which uses wgpu::Backends::all() and wgpu::Dx12Compiler::Fxc for InstanceDescriptor
- let instance = Instance::new(wgpu::InstanceDescriptor {
- backends: wgpu::Backends::all(),
- dx12_shader_compiler: wgpu::Dx12Compiler::Fxc,
- });
+ let instance = Instance::default();By @Elabajaba in #3356
The new view_formats field in the TextureDescriptor is used to specify a list of formats the texture can be re-interpreted to in a texture view. Currently only changing srgb-ness is allowed (ex. Rgba8Unorm <=> Rgba8UnormSrgb).
let texture = device.create_texture(&wgpu::TextureDescriptor {
// ...
format: TextureFormat::Rgba8UnormSrgb,
+ view_formats: &[TextureFormat::Rgba8Unorm],
});let config = wgpu::SurfaceConfiguration {
// ...
format: TextureFormat::Rgba8Unorm,
+ view_formats: vec![wgpu::TextureFormat::Rgba8UnormSrgb],
};
surface.configure(&device, &config);Via the TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURES feature, MSAA x2 and x8 are now supported on textures. To query for x2 or x8 support, enable the feature and look at the texture format flags for the texture format of your choice.
By @39ali in 3140
You can now choose to use the DXC compiler for DX12 instead of FXC. The DXC compiler is faster, less buggy, and allows for new features compared to the old, unmaintained FXC compiler.
You can choose which compiler to use at Instance creation using the dx12_shader_compiler field in the InstanceDescriptor struct. Note that DXC requires both dxcompiler.dll and dxil.dll, which can be downloaded from https://github.com/microsoft/DirectXShaderCompiler/releases. Both .dlls need to be shipped with your application when targeting DX12 and using the DXC compiler. If the .dlls can't be loaded, then it will fall back to the FXC compiler. By @39ali and @Elabajaba in #3356
The DX12 backend can now suballocate buffers and textures from larger chunks of memory, which can give a significant increase in performance (in testing a 100x improvement has been seen in a simple scene with 200 write_buffer calls per frame, and a 1.4x improvement in Bistro using Bevy).
Previously wgpu-hal's DX12 backend created a new heap on the GPU every time you called write_buffer (by calling CreateCommittedResource), whereas now it uses gpu_allocator to manage GPU memory (and calls CreatePlacedResource with a suballocated heap). By @Elabajaba in #3163
Whereas wgpu-core used to automatically select backends to enable
based on the target OS and architecture, it now has separate features
to enable each backend:
- "metal", for the Metal API on macOS and iOS
- "vulkan", for the Vulkan API (Linux, some Android, and occasionally Windows)
- "dx12", for Microsoft's Direct3D 12 API
- "gles", OpenGL ES, available on many systems
- "dx11", for Microsoft's Direct3D 11 API
None are enabled by default, but the wgpu crate automatically
selects these features based on the target operating system and
architecture, using the same rules that wgpu-core used to, so users
of wgpu should be unaffected by this change. However, other crates
using wgpu-core directly will need to copy wgpu's logic or write
their own. See the [target] section of wgpu/Cargo.toml for
details.
Similarly, wgpu-core now has emscripten and renderdoc features
that wgpu enables on appropriate platforms.
In previous releases, the wgpu-core crate decided which backends to
support. However, this left wgpu-core's users with no way to
override those choices. (Firefox doesn't want the GLES back end, for
example.) There doesn't seem to be any way to have a crate select
backends based on target OS and architecture that users of that crate
can still override. Default features can't be selected based on the
target, for example. That implies that we should do the selection as
late in the dependency DAG as feasible. Having wgpu (and
wgpu-core's other dependents) choose backends seems like the best
option.
By @jimblandy in #3254.
- Convert all
DefaultImplementations on Enums toderive(Default) - Implement
DefaultforCompositeAlphaMode - New downlevel feature
UNRESTRICTED_INDEX_BUFFERto indicate support for usingINDEXtogether with other non-copy/map usages (unsupported on WebGL). By @Wumpf in #3157 - Add missing
DEPTH_BIAS_CLAMPandFULL_DRAW_INDEX_UINT32downlevel flags. By @teoxoy in #3316 - Combine
Surface::get_supported_formats,Surface::get_supported_present_modes, andSurface::get_supported_alpha_modesintoSurface::get_capabilitiesandSurfaceCapabilities. By @cwfitzgerald in #3157 - Make
Surface::get_default_configreturn an Option to prevent panics. By @cwfitzgerald in #3157 - Lower the
max_buffer_sizelimit value for compatibility with Apple2 and WebGPU compliance. By @jinleili in #3255 - Limits
min_uniform_buffer_offset_alignmentandmin_storage_buffer_offset_alignmentis now always at least 32. By @wumpf #3262 - Dereferencing a buffer view is now marked inline. By @Wumpf in #3307
- The
strict_assertfamily of macros was moved towgpu-types. By @i509VCB in #3051 - Make
ObjectIdstructure and invariants idiomatic. By @teoxoy in #3347 - Add validation in accordance with WebGPU
GPUSamplerDescriptorvalid usage forlodMinClampandlodMaxClamp. By @James2022-rgb in #3353 - Remove panics in
Derefimplementations forQueueWriteBufferViewandBufferViewMut. Instead, warnings are logged, since reading from these types is not recommended. By @botahamec in [#3336] - Implement
view_formatsin the TextureDescriptor to match the WebGPU spec. By @jinleili in #3237 - Show more information in error message for non-aligned buffer bindings in WebGL #3414
- Update
TextureViewvalidation according to the WebGPU spec. By @teoxoy in #3410 - Implement
view_formatsin the SurfaceConfiguration to match the WebGPU spec. By @jinleili in #3409
- Set
WEBGPU_TEXTURE_FORMAT_SUPPORTdownlevel flag depending on the proper format support by @teoxoy in #3367. - Set
COPY_SRC/COPY_DSTonly based on Vulkan'sTRANSFER_SRC/TRANSFER_DSTby @teoxoy in #3366.
- Browsers that support
OVR_multiview2now report theMULTIVIEWfeature by @expenses in #3121. Limits::max_push_constant_sizeon GLES is now 256 by @Dinnerbone in #3374.- Creating multiple pipelines with the same shaders will now be faster, by @Dinnerbone in #3380.
- Implement
queue_validate_write_bufferby @jinleili in #3098 - Sync depth/stencil copy restrictions with the spec by @teoxoy in #3314
- Implement
HashforDepthStencilStateandDepthBiasState - Add the
"wgsl"feature, to enable WGSL shaders inwgpu-coreandwgpu. Enabled by default inwgpu. By @daxpedda in #2890. - Implement
CloneforShaderSourceandShaderModuleDescriptorinwgpu. By @daxpedda in #3086. - Add
get_default_configforSurfaceto simplify user creation ofSurfaceConfiguration. By @jinleili in #3034 - Improve compute shader validation error message. By @haraldreingruber in #3139
- Native adapters can now use MSAA x2 and x8 if it's supported , previously only x1 and x4 were supported . By @39ali in 3140
- Implemented correleation between user timestamps and platform specific presentation timestamps via [
Adapter::get_presentation_timestamp]. By @cwfitzgerald in #3240 - Added support for
Features::SHADER_PRIMITIVE_INDEXon all backends. By @cwfitzgerald in #3272 - Implemented
TextureFormat::Stencil8, allowing for stencil testing without depth components. By @Dinnerbone in #3343 - Implemented
add_srgb_suffix()forTextureFormatfor converting linear formats to sRGB. By @Elabajaba in #3419 - Zero-initialize workgroup memory. By @teoxoy in #3174
- Surfaces support now
TextureFormat::Rgba8Unormand (non-web only)TextureFormat::Bgra8Unorm. By @Wumpf in #3070 - Support alpha to coverage. By @Wumpf in #3156
- Support filtering f32 textures. By @expenses in #3261
- Add
SHADER_INT16feature to enable theshaderInt16VkPhysicalDeviceFeature. By @Elabajaba in #3401
- Add
MULTISAMPLE_X2,MULTISAMPLE_X4andMULTISAMPLE_X8toTextureFormatFeatureFlags. By @39ali in 3140 - Sync
TextureFormat.describewith the spec. By @teoxoy in 3312
- Add a way to create
DeviceandQueuefrom raw Metal resources in wgpu-hal. By @AdrianEddy in #3338
- Update ndk-sys to v0.4.1+23.1.7779620, to fix checksum failures. By @jimblandy in #3232.
- Bother to free the
hal::Api::CommandBufferwhen awgpu_core::command::CommandEncoderis dropped. By @jimblandy in #3069. - Fixed the mipmap example by adding the missing WRITE_TIMESTAMP_INSIDE_PASSES feature. By @Olaroll in #3081.
- Avoid panicking in some interactions with invalid resources by @nical in (#3094)[#3094]
- Fixed an integer overflow in
copy_texture_to_textureby @nical #3090 - Remove
wgpu_types::Features::DEPTH24PLUS_STENCIL8, makingwgpu::TextureFormat::Depth24PlusStencil8available on all backends. By @Healthire in (#3151)[#3151] - Fix an integer overflow in
queue_write_textureby @nical in (#3146)[#3146] - Make
RenderPassCompatibilityErrorandCreateShaderModuleErrornot so huge. By @jimblandy in (#3226)[#3226] - Check for invalid bitflag bits in wgpu-core and allow them to be captured/replayed by @nical in (#3229)[#3229]
- Evaluate
gfx_select!'s#[cfg]conditions at the right time. By @jimblandy in #3253 - Improve error messages when binding bind group with dynamic offsets. By @cwfitzgerald in #3294
- Allow non-filtering sampling of integer textures. By @JMS55 in #3362.
- Validate texture ids in
Global::queue_texture_write. By @jimblandy in #3378. - Don't panic on mapped buffer in queue_submit. By @crowlKats in #3364.
- Fix being able to sample a depth texture with a filtering sampler. By @teoxoy in #3394.
- Make
make_spirv_rawandmake_spirvhandle big-endian binaries. By @1e1001 in #3411.
- Update ash to 0.37.1+1.3.235 to fix CI breaking by changing a call to the deprecated
debug_utils_set_object_name()function toset_debug_utils_object_name()by @elabajaba in #3273 - Document and improve extension detection. By @teoxoy in #3327
- Don't use a pointer to a local copy of a
PhysicalDeviceDriverPropertiesstruct after it has gone out of scope. In fact, don't make a local copy at all. Introduce a helper function for buildingCStrs from C character arrays, and remove someunsafeblocks. By @jimblandy in #3076.
- Fix
depth16Unormformats by @teoxoy in #3313 - Don't re-use
GraphicsCommandListwhencloseorresetfails. By @xiaopengli89 in #3204
- Fix texture view creation with full-resource views when using an explicit
mip_level_countorarray_layer_count. By @cwfitzgerald in #3323
- Fixed WebGL not displaying srgb targets correctly if a non-screen filling viewport was previously set. By @Wumpf in #3093
- Fix disallowing multisampling for float textures if otherwise supported. By @Wumpf in #3183
- Fix a panic when creating a pipeline with opaque types other than samplers (images and atomic counters). By @James2022-rgb in #3361
- Fix uniform buffers being empty on some vendors. By @Dinnerbone in #3391
- Fix a panic allocating a new buffer on webgl. By @Dinnerbone in #3396
- Use
loginstead ofprintlnin hello example by @JolifantoBambla in #2858
- Let
setVertexBufferandsetIndexBuffercalls onGPURenderBundleEncoderthrow an error if thesizeargument is zero, rather than treating that as "until the end of the buffer". By @jimblandy in #3171
- Let the wgpu examples
framework.rscompile again under Emscripten. By @jimblandy in #3246
- Log adapter info in hello example on wasm target by @JolifantoBambla in #2858
- Added new example
stencil-trianglesto show basic use of stencil testing. By @Dinnerbone in #3343
- Update the
minimum supported rust versionto 1.64 - Move
ResourceMetadatainto its own module. By @jimblandy in #3213 - Add WebAssembly testing infrastructure. By @haraldreingruber in #3238
- Error message when you forget to use cargo-nextest. By @cwfitzgerald in #3293
- Fix all suggestions from
cargo clippy
- Fix incorrect offset in
get_mapped_rangeby @nical in #3233
- Make
wgpu::TextureFormat::Depth24PlusStencil8available on all backends by making the feature unconditionally available and the feature unneeded to use the format. By @Healthire and @cwfitzgerald in #3165
When using CompareFunction::Equal or CompareFunction::NotEqual on a pipeline, there is now a warning logged if the vertex shader does not have a @invariant tag on it. On some machines, rendering the same triangles multiple times without an @invariant tag will result in slightly different depths for every pixel. Because the *Equal functions rely on depth being the same every time it is rendered, we now warn if it is missing.
-@vertex
-fn vert_main(v_in: VertexInput) -> @builtin(position) vec4<f32> {...}
+@vertex
+fn vert_main(v_in: VertexInput) -> @builtin(position) @invariant vec4<f32> {...}Surface supports alpha_mode now. When alpha_mode is equal to PreMultiplied or PostMultiplied,
the alpha channel of framebuffer is respected in the compositing process, but which mode is available depends on
the different API and Device. If don't care about alpha_mode, you can set it to Auto.
SurfaceConfiguration {
// ...
+ alpha_mode: surface.get_supported_alpha_modes(&adapter)[0],
}The function to enumerate supported presentation modes changed:
- pub fn wgpu::Surface::get_supported_modes(&self, adapter: &wgpu::Adapter) -> Vec<PresentMode>
+ pub fn wgpu::Surface::get_supported_present_modes(&self, adapter: &wgpu::Adapter) -> Vec<PresentMode>This will allow use of the latest version of winit. As such the bound on create_surface is now RWH 0.5 and requires
both raw_window_handle::HasRawWindowHandle and raw_window_handle::HasRawDisplayHandle.
- Add
Buffer::size()andBuffer::usage(); by @kpreid in #2923 - Split Blendability and Filterability into Two Different TextureFormatFeatureFlags; by @stakka in #3012
- Expose
alpha_modeon SurfaceConfiguration, by @jinleili in #2836 - Introduce fields for driver name and info in
AdapterInfo, by @i509VCB in #3037 - Add way to create gles hal textures from raw gl names to allow externally managed textures. By @i509VCB #3046
- Implemented
copy_external_image_to_textureon WebGPU, by @ybiletskyi in #2781
- Free
StagingBufferseven when an error occurs in the operation that consumes them. By @jimblandy in #2961 - Avoid overflow when checking that texture copies fall within bounds. By @jimblandy in #2963
- Improve the validation and error reporting of buffer mappings by @nical in #2848
- Fix compilation errors when using wgpu-core in isolation while targeting
wasm32-unknown-unknownby @Seamooo in #2922 - Fixed opening of RenderDoc library by @abuffseagull in #2930
- Added missing validation for
BufferUsagesmismatches whenFeatures::MAPPABLE_PRIMARY_BUFFERSis not enabled. By @imberflur in #3023 - Fixed
CommandEncodernot beingSendandSyncon web by @i509VCB in #3025 - Document meaning of
vendorinAdapterInfoif the vendor has no PCI id. - Fix missing resource labels from some Errors by @scoopr in #3066
- Add the missing
msg_send![view, retain]call withinfrom_viewby @jinleili in #2976 - Fix
max_buffermax_textureandmax_vertex_bufferslimits by @jinleili in #2978 - Remove PrivateCapabilities's
format_rgb10a2_unorm_surfacefield by @jinleili in #2981 - Fix validation error when copying into a subset of a single-layer texture by @nical in #3063
- Fix
_buffer_sizesencoding by @dtiselice in #3047
- Fix
astc_hdrformats support by @jinleili in [#2971]](#2971) - Update to Naga b209d911 (2022-9-1) to avoid generating SPIR-V that
violates Vulkan valid usage rules
VUID-StandaloneSpirv-Flat-06202andVUID-StandaloneSpirv-Flat-04744. By @jimblandy in #3008 - Fix bug where the Vulkan backend would panic when using a supported window and display handle but the dependent extensions are not available by @i509VCB in #3054.
- Report vendor id for Mesa and Apple GPUs. By @i509VCB #3036
- Report Apple M2 gpu as integrated. By @i509VCB #3036
- When called in a web worker,
Context::init()now usesweb_sys::WorkerGlobalContextto create awgpu::Instanceinstead of trying to access the unavailableweb_sys::Windowby @JolifantoBambla in #2858
- Changed wgpu-hal and wgpu-core implementation to pass RawDisplayHandle and RawWindowHandle as separate parameters instead of passing an impl trait over both HasRawDisplayHandle and HasRawWindowHandle. By @i509VCB in #3022
- Changed
Instance::as_hal<A>to just return anOption<&A::Instance>rather than taking a callback. By @jimb in #2991 - Added downlevel restriction error message for
InvalidFormatUsageserror by @Seamooo in #2886 - Add warning when using CompareFunction::*Equal with vertex shader that is missing @invariant tag by @cwfitzgerald in #2887
- Update Winit to version 0.27 and raw-window-handle to 0.5 by @wyatt-herkamp in #2918
- Address Clippy 0.1.63 complaints. By @jimblandy in #2977
- Don't use
PhantomDataforIdentityManager'sInputtype. By @jimblandy in #2972 - Changed Naga variant in ShaderSource to
Cow<'static, Module>, to allow loading global variables by @daxpedda in #2903 - Updated the maximum binding index to match the WebGPU specification by @nical in #2957
- Add
unsafe_op_in_unsafe_fnto Clippy lints in the entire workspace. By @ErichDonGubler in #3044.
- Extract the generic code into
get_metal_layerby @jinleili in #2826
- Remove use of Vulkan12Features/Properties types. By @i509VCB in #2936
- Provide a means for
wgpuusers to accessvk::Queueand the queue index. By @anlumo in #2950 - Use the use effective api version for determining device features instead of wrongly assuming
VkPhysicalDeviceProperties.apiVersionis the actual version of the device. By @i509VCB in #3011 DropGuardhas been moved to the root of the wgpu-hal crate. By @i509VCB #3046
- Add
Rgba16Floatformat support for color attachments. By @jinleili in #3045 TEXTURE_COMPRESSION_ASTC_HDRfeature detection by @jinleili in #3042
- Made
StagingBelt::write_buffer()check more thoroughly for reusable memory; by @kpreid in #2906
- Add WGSL examples to complement existing examples written in GLSL by @norepimorphism in #2888
- Document
wgpu_coreresource allocation. @jimblandy in #2973 - Expanded
StagingBeltdocumentation by @kpreid in #2905 - Fixed documentation for
Instance::create_surface_from_canvasandInstance::create_surface_from_offscreen_canvasregarding their safety contract. These functions are not unsafe. By @jimblandy #2990 - Document that
write_buffer_with()is sound but unwise to read from by @kpreid in #3006 - Explain why
Adapter::as_halandDevice::as_halhave to take callback functions. By @jimblandy in #2992
- Update wasm32 dependencies, set
alpha_modeon web target by @jinleili in #3040
- Add the
"strict_asserts"feature, to enable additional internal run-time validation inwgpu-core. By @jimblandy in #2872.
Manual concatenation of cargo public-api --diff-git-checkouts v0.13.2 v0.14.0 -p wgpu and cargo public-api --diff-git-checkouts v0.13.2 v0.14.0 -p wgpu-types
Removed items from the public API
=================================
-pub fn wgpu::Surface::get_supported_modes(&self, adapter: &wgpu::Adapter) -> Vec<PresentMode>
-pub const wgpu::Features::DEPTH24UNORM_STENCIL8: Self
-pub enum variant wgpu::TextureFormat::Depth24UnormStencil8
Changed items in the public API
===============================
-pub unsafe fn wgpu::Instance::as_hal<A: wgc::hub::HalApi, F: FnOnce(Option<&<A as >::Instance>) -> R, R>(&self, hal_instance_callback: F) -> R
+pub unsafe fn wgpu::Instance::as_hal<A: wgc::hub::HalApi>(&self) -> Option<&<A as >::Instance>
-pub unsafe fn wgpu::Instance::create_surface<W: raw_window_handle::HasRawWindowHandle>(&self, window: &W) -> wgpu::Surface
+pub unsafe fn wgpu::Instance::create_surface<W: raw_window_handle::HasRawWindowHandle + raw_window_handle::HasRawDisplayHandle>(&self, window: &W) -> wgpu::Surface
Added items to the public API
=============================
+pub fn wgpu::Buffer::size(&self) -> wgt::BufferAddress
+pub fn wgpu::Buffer::usage(&self) -> BufferUsages
+pub fn wgpu::Surface::get_supported_alpha_modes(&self, adapter: &wgpu::Adapter) -> Vec<CompositeAlphaMode>
+pub fn wgpu::Surface::get_supported_present_modes(&self, adapter: &wgpu::Adapter) -> Vec<PresentMode>
+#[repr(C)] pub enum wgpu::CompositeAlphaMode
+impl RefUnwindSafe for wgpu::CompositeAlphaMode
+impl Send for wgpu::CompositeAlphaMode
+impl Sync for wgpu::CompositeAlphaMode
+impl Unpin for wgpu::CompositeAlphaMode
+impl UnwindSafe for wgpu::CompositeAlphaMode
+pub const wgpu::Features::DEPTH24PLUS_STENCIL8: Self
+pub const wgpu::TextureFormatFeatureFlags::BLENDABLE: Self
+pub enum variant wgpu::CompositeAlphaMode::Auto = 0
+pub enum variant wgpu::CompositeAlphaMode::Inherit = 4
+pub enum variant wgpu::CompositeAlphaMode::Opaque = 1
+pub enum variant wgpu::CompositeAlphaMode::PostMultiplied = 3
+pub enum variant wgpu::CompositeAlphaMode::PreMultiplied = 2
+pub enum variant wgpu::TextureFormat::Depth16Unorm
+pub fn wgpu::CompositeAlphaMode::clone(&self) -> wgpu::CompositeAlphaMode
+pub fn wgpu::CompositeAlphaMode::eq(&self, other: &wgpu::CompositeAlphaMode) -> bool
+pub fn wgpu::CompositeAlphaMode::fmt(&self, f: &mut $crate::fmt::Formatter<'_>) -> $crate::fmt::Result
+pub fn wgpu::CompositeAlphaMode::hash<__H: $crate::hash::Hasher>(&self, state: &mut __H) -> ()
+pub struct field wgpu::AdapterInfo::driver: String
+pub struct field wgpu::AdapterInfo::driver_info: String
+pub struct field wgpu::SurfaceConfiguration::alpha_mode: wgpu_types::CompositeAlphaMode- Prefer
DeviceType::DiscreteGpuoverDeviceType::OtherforPowerPreference::LowPowerso Vulkan is preferred over OpenGL again by @Craig-Macomber in #2853 - Allow running
get_texture_format_featureson unsupported texture formats (returning no flags) by @cwfitzgerald in #2856 - Allow multi-sampled textures that are supported by the device but not WebGPU if
TEXTURE_ADAPTER_SPECIFIC_FORMAT_FEATURESis enabled by @cwfitzgerald in #2856 get_texture_format_featuresonly lists the COPY_* usages if the adapter actually supports that usage by @cwfitzgerald in #2856- Fix bind group / pipeline deduplication not taking into account RenderBundle execution resetting these values by @shoebe #2867
- Fix panics that occur when using
as_halfunctions when the hal generic type does not match the hub being looked up in by @i509VCB #2871 - Add some validation in map_async by @nical in #2876
- Fix bugs when mapping/unmapping zero-sized buffers and ranges by @nical in #2877
- Fix out-of-bound write in
map_bufferwith non-zero offset by @nical in #2916 - Validate the number of color attachments in
create_render_pipelineby @nical in #2913 - Validate against the maximum binding index in
create_bind_group_layoutby @nical in #2892 - Validate that map_async's range is not negative by @nical in #2938
- Fix calculation/validation of layer/mip ranges in create_texture_view by @nical in #2955
- Validate the sample count and mip level in
copy_texture_to_bufferby @nical in #2958 - Expose the cause of the error in the
map_asynccallback in #2939
DownlevelCapabilities::default()now returns theANISOTROPIC_FILTERINGflag set to true so DX12 listsANISOTROPIC_FILTERINGas true again by @cwfitzgerald in #2851- Properly query format features for UAV/SRV usages of depth formats by @cwfitzgerald in #2856
- Vulkan 1.0 drivers that support
VK_KHR_multiviewnow properly report theMULTIVIEWfeature as supported by @i509VCB in #2934. - Stop using
VkPhysicalDevice11Featuresin Vulkan 1.1 which is confusingly provided in Vulkan 1.2 by @i509VCB in #2934.
- Fix depth stencil texture format capability by @jinleili in #2854
get_texture_format_featuresnow only returns usages for formats it actually supports by @cwfitzgerald in #2856
- Allow access to queue family index in Vulkan hal by @i509VCB in #2859
- Allow access to the EGLDisplay and EGLContext pointer in Gles hal Adapter and Device by @i509VCB in #2860
- Update present_mode docs as most of them don't automatically fall back to Fifo anymore. by @Elabajaba in #2855
- Document safety requirements for
Adapter::from_externalin gles hal by @i509VCB in #2863 - Make
AdapterContexta publicly accessible type in the gles hal by @i509VCB in #2870
- Fix out of bounds access when surface texture is written to by multiple command buffers by @cwfitzgerald in #2843
- AutoNoVSync now correctly falls back to Fifo by @simbleau in #2842
- Fix GL_EXT_color_buffer_float detection on native by @cwfitzgerald in #2843
WGSL syntax has changed in a couple ways. The new syntax is easier to read and work with.
Attribute declarations are written differently:
- [[group(1), binding(0)]]
+ @group(1) @binding(0)Stage declarations are now separate attributes rather than part of the stage attribute:
- [[stage(vertex)]]
+ @vertexStructs now use , as field separator and no longer need semicolons after the declaration:
- struct MyStruct {
- my_member: u32;
- };
+ struct MyStruct {
+ my_member: u32,
+ }The method of getting the preferred swapchain format has changed to allow viewing all formats supported by the surface.
- let format = surface.get_preferred_format(&adapter).unwrap();
+ let format = surface.get_supported_formats(&adapter)[0];Presentation modes now need to match exactly what the surface supports. FIFO is always supported,
but all other modes vary from API to API and Device to Device. To get a list of all supported modes,
call the following. The order does not indicate preference.
let modes = surface.get_supported_present_modes(&adapter);Timestamp queries are now restricted behind multiple features to allow implementation on TBDR (Tile-Based Deferred Rendering) based GPUs, such as mobile devices and Apple's M chips.
Features::TIMESTAMP_QUERIES now allows for calling write_timestamp only on CommandEncoders.
Features::WRITE_TIMESTAMP_INSIDE_PASSES is needed to call write_timestamp on RenderPassEncoders or ComputePassEncoders.
The function for mapping buffers no longer returns a future, and instead calls a callback when the buffer is mapped.
This aligns with the use of the API more clearly - you aren't supposed to block and wait on the future to resolve, you are supposed to keep rendering and wait until the buffer maps on its own. Mapping and the flow of mapping is an under-documented area that we hope to improve in the future.
- let future = buffer.slice(..).map_async(MapMode::Read);
+ buffer.slice(..).map_async(MapMode::Read, || {
+ // Called when buffer is mapped.
+ })Calling queue.submit now returns an opaque submission index that can be used as an argument to
device.poll to say which submission to wait to complete.
Device::create_shader_module now takes the shader descriptor by value:
- device.create_shader_module(&shader_module_descriptor)
+ device.create_shader_module(shader_module_descriptor)Color attachments can be sparse, so they are now optional:
FragmentState {
- targets: &[color_target_state]
+ targets: &[Some(color_target_state)]
// ..
}RenderPassDescriptor {
- color_attachments: &[render_pass_color_attachment]
+ color_attachments: &[Some(render_pass_color_attachment)]
// ..
}RenderBundleEncoderDescriptor {
- color_formats: &[texture_format]
+ color_formats: &[Some(texture_format)]
// ..
}Extent3d::max_mips now requires you to pass a TextureDimension to specify whether or not depth_or_array_layers should be ignored:
Extent3d {
width: 1920,
height: 1080,
depth_or_array_layers: 6,
- }.max_mips()
+ }.max_mips(wgpu::TextureDimension::D3)Limits has a new field, max_buffer_size (not an issue if you don't define limits manually):
Limits {
// ...
+ max_buffer_size: 256 * 1024 * 1024, // adjust as you see fit
}Features::CLEAR_COMMANDS is now unnecessary and no longer exists. The feature to clear buffers and textures is now part of upstream WebGPU.
DeviceDescriptor {
// ...
features: wgpu::Features::VERTEX_WRITABLE_STORAGE
| wgpu::Features::MAPPABLE_PRIMARY_BUFFERS
| wgpu::Features::TEXTURE_BINDING_ARRAY
| wgpu::Features::BUFFER_BINDING_ARRAY
| wgpu::Features::STORAGE_RESOURCE_BINDING_ARRAY
- | wgpu::Features::CLEAR_COMMANDS
,
}ComputePass::dispatch has been renamed to ComputePass::dispatch_workgroups
- cpass.dispatch(self.work_group_count, 1, 1)
+ cpass.dispatch_workgroups(self.work_group_count, 1, 1)- Add
util::indirect::*helper structs by @IcanDivideBy0 in #2365 - Add
AddressMode::ClampToZeroby @laptou in #2364 - Add MULTISAMPLED_SHADING downlevel flag by @jinleili in #2425
- Allow non struct buffers in wgsl by @IcanDivideBy0 in #2451
- Prefix every wgpu-generated label with
(wgpu). by @kpreid in #2590 - Permit non-struct, non-array types as buffers. by @jimblandy in #2584
- Return
queue_emptyfor Device::poll by @xiaopengli89 in #2643 - Add
SHADER_FLOAT16feature by @jinleili in #2646 - Add DEPTH32FLOAT_STENCIL8 feature by @jinleili in #2664
- Add DEPTH24UNORM_STENCIL8 feature by @jinleili in #2689
- Implement submission indexes by @cwfitzgerald in #2700
- [WebGL] Add a downlevel capability for rendering to floating point textures by @expenses in #2729
- allow creating wgpu::Instance from wgpu_core::Instance by @i509VCB in #2763
- Force binding sizes to be multiples of 16 on webgl by @cwfitzgerald in #2808
- Add Naga variant to ShaderSource by @rttad in #2801
- Implement Queue::write_buffer_with by @teoxoy in #2777
- Re-allow vk backend on Apple platforms via
vulkan-portabilityfeature by @jinleili in #2488 - vulkan: HDR ASTC formats support by @jinleili in #2496
- Implement push constants for metal backend by @TheOnlyMrCat in #2314
- Metal backend ASTC HDR formats support by @jinleili in #2477
- Add COPY_DST to Metal's surface usage bits by @vl4dimir in #2491
- Add
Features::MULTI_DRAW_INDIRECTto Metal by @expenses in #2737
- Support externally initialized contexts by @kvark in #2350
- Angle support on macOS by @jinleili in #2461
- Use EGL surfaceless platform when windowing system is not found by @sh7dm in #2339
- Do a downlevel check for anisotrophy and enable it in the webgl backend by @expenses in #2616
- OffscreenCanvas Support for WebGL Backend by @haraldreingruber-dedalus in #2603
- Support to create surface from visual on Windows by @xiaopengli89 in #2434
- Add raw_queue for d3d12 device by @xiaopengli89 in #2600
- Skeleton of a DX11 backend - not working yet by @cwfitzgerald in #2443
- Adapter and Instance as_hal functions by @i509VCB in #2663
- expose some underlying types in Vulkan hal by @i509VCB in #2667
- Add raw_device method for dx12, vulkan hal by @xiaopengli89 in #2360
- expose egl display in gles Instance hal by @i509VCB in #2670
- Add raw_adapter method for dx12 hal adapter by @xiaopengli89 in #2714
- Acquire texture:
Option<std::time::Duration>timeouts by @rib in #2724 - expose vulkan physical device capabilities, enabled device extensions by @i509VCB in #2688
- feature: emscripten by @caiiiycuk in #2422
- feature = emscripten, compatibility fixes for wgpu-native by @caiiiycuk in #2450
- Make ShaderSource #[non_exhaustive] by @fintelia in #2312
- Make
execute_bundles()receive IntoIterator by @maku693 in #2410 - Raise
wgpu_hal::MAX_COLOR_TARGETSto 8. by @jimblandy in #2640 - Rename dispatch -> dispatch_workgroups by @jinleili in #2619
- Update texture_create_view logic to match spec by @jinleili in #2621
- Move TEXTURE_COMPRESSION_ETC2 | ASTC_LDR to web section to match spec by @jinleili in #2671
- Check that all vertex outputs are consumed by the fragment shader by @cwfitzgerald in #2704
- Convert map_async from being async to being callback based by @cwfitzgerald in #2698
- Align the validation of Device::create_texture with the WebGPU spec by @nical in #2759
- Add InvalidGroupIndex validation at create_shader_module by @jinleili in #2775
- Rename MAX_COLOR_TARGETS to MAX_COLOR_ATTACHMENTS to match spec by @jinleili in #2780
- Change get_preferred_format to get_supported_formats by @stevenhuyn in #2783
- Restrict WriteTimestamp Inside Passes by @cwfitzgerald in #2802
- Flip span labels to work better with tools by @cwfitzgerald in #2820
- Make GLES DeviceType unknown by default by @PolyMeilex in #2647
- metal: check if in the main thread when calling
create_surfaceby @jinleili in #2736
- limit binding sizes to i32 by @kvark in #2363
- Fix trac(y/ing) compile issue by @cwfitzgerald in #2333
- Improve detection and validation of cubemap views by @kvark in #2331
- Don't create array layer trackers for 3D textures. by @ElectronicRU in #2348
- Limit 1D texture mips to 1 by @kvark in #2374
- Texture format MSAA capabilities by @kvark in #2377
- Fix write_buffer to surface texture @kvark in #2385
- Improve some error messages by @cwfitzgerald in #2446
- Don't recycle indices that reach EOL by @kvark in #2462
- Validated render usages for 3D textures by @kvark in #2482
- Wrap all validation logs with catch_unwinds by @cwfitzgerald in #2511
- Fix clippy lints by @a1phyr in #2560
- Free the raw device when
wgpu::Deviceis dropped. by @jimblandy in #2567 - wgpu-core: Register new pipelines with device's tracker. by @jimblandy in #2565
- impl Debug for StagingBelt by @kpreid in #2572
- Use fully qualified syntax for some calls. by @jimblandy in #2655
- fix: panic in
Storage::getby @SparkyPotato in #2657 - Report invalid pipelines in render bundles as errors, not panics. by @jimblandy in #2666
- Perform "valid to use with" checks when recording render bundles. by @jimblandy in #2690
- Stop using storage usage for sampling by @cwfitzgerald in #2703
- Track depth and stencil writability separately. by @jimblandy in #2693
- Improve InvalidScissorRect error message by @jinleili in #2713
- Improve InvalidViewport error message by @jinleili in #2723
- Don't dirty the vertex buffer for stride/rate changes on bundles. by @jimblandy in #2744
- Clean up render bundle index buffer tracking. by @jimblandy in #2743
- Improve read-write and read-only texture storage error message by @jinleili in #2745
- Change
WEBGPU_TEXTURE_FORMAT_SUPPORTto1 << 14instead of1 << 15by @expenses in #2772 - fix BufferMapCallbackC & SubmittedWorkDoneClosureC by @rajveermalviya in #2787
- Fix formatting of
TextureDimensionError::LimitExceeded. by @kpreid in #2799 - Remove redundant
#[cfg]conditions frombackend/direct.rs. by @jimblandy in #2811 - Replace android-properties with android_system_properties. by @nical in #2815
- Relax render pass color_attachments validation by @jinleili in #2778
- Properly Barrier Compute Indirect Buffers by @cwfitzgerald in #2810
- Use numeric constants to define
wgpu_types::Featuresvalues. by @jimblandy in #2817
- Fix surface texture clear view by @kvark in #2341
- Set preserveInvariance for shader options by @scoopr in #2372
- Properly set msl version to 2.3 if supported by @cwfitzgerald in #2418
- Identify Apple M1 GPU as integrated by @superdump in #2429
- Fix M1 in macOS incorrectly reports supported compressed texture formats by @superdump in #2453
- Msl: support unsized array not in structures by @kvark in #2459
- Fix
Surface::from_uiviewcan not guarantee set correctcontentScaleFactorby @jinleili in #2470 - Set
max_buffer_sizeby the correct physical device restriction by @jinleili in #2502 - Refactor
PrivateCapabilitiescreation by @jinleili in #2509 - Refactor texture_format_capabilities function by @jinleili in #2522
- Improve
push | pop_debug_markerby @jinleili in #2537 - Fix some supported limits by @jinleili in #2608
- Don't skip incomplete binding resources. by @dragostis in #2622
- Fix
Rgb9e5Ufloatcapabilities andsampler_lod_averagesupport by @jinleili in #2656 - Fix Depth24Plus | Depth24PlusStencil8 capabilities by @jinleili in #2686
- Get_supported_formats: sort like the old get_preferred_format and simplify return type by @victorvde in #2786
- Restrict hal::TextureUses::COLOR_TARGET condition within create_texture by @jinleili in #2818
- Fix UMA check by @kvark in #2305
- Fix partial texture barrier not affecting stencil aspect by @Wumpf in #2308
- Improve RowPitch computation by @kvark in #2409
- Explicitly set Vulkan debug message types instead of !empty() by @victorvde in #2321
- Use stencil read/write masks by @kvark in #2382
- Vulkan: correctly set INDEPENDENT_BLEND，make runable on Android 8.x by @jinleili in #2498
- Fix ASTC format mapping by @kvark in #2476
- Support flipped Y on VK 1.1 devices by @cwfitzgerald in #2512
- Fixed builtin(primitive_index) for vulkan backend by @kwillemsen in #2716
- Fix PIPELINE_STATISTICS_QUERY feature support by @jinleili in #2750
- Add a vulkan workaround for large buffers. by @nical in #2796
- Fix index buffer state not being reset in reset_state by @rparrett in #2391
- Allow push constants trough emulation by @JCapucho in #2400
- Hal/gles: fix dirty vertex buffers that are unused by @kvark in #2427
- Fix texture description for bgra formats by @JCapucho in #2520
- Remove a
log::error!debugging statement from the gles queue by @expenses in #2630 - Fix clearing depth and stencil at the same time by @expenses in #2675
- Handle cubemap copies by @expenses in #2725
- Allow clearing index buffers by @grovesNL in #2740
- Fix buffer-texture copy for 2d arrays by @tuchs in #2809
- Search for different versions of libwayland by @sh7dm in #2336
- Fix compilation on wasm32-unknown-unknown without
webglfeature by @jakobhellermann in #2355 - Solve crash on WebGPU by @cwfitzgerald in #2807
- Fix emscripten by @cwfitzgerald in #2494
- Do texture init via clear passes when possible by @Wumpf in #2307
- Bind group deduplication by @cwfitzgerald in #2623
- Tracking Optimization and Rewrite by @cwfitzgerald in #2662
- Add defaults to new limits and correct older ones by @MultisampledNight in #/2303
- Improve shader source documentation by @grovesNL in #2315
- Fix typo by @rustui in #2393
- Add a ⭐ to the feature matrix of examples README by @yutannihilation in #2457
- Fix get_timestamp_period type in docs by @superdump in #2478
- Fix mistake in Access doc comment by @nical in #2479
- Improve shader support documentation by @cwfitzgerald in #2501
- Document the gfx_select! macro. by @jimblandy in #2555
- Add Windows 11 to section about DX12 by @HeavyRain266 in #2552
- Document some aspects of resource tracking. by @jimblandy in #2558
- Documentation for various things. by @jimblandy in #2566
- Fix doc links. by @jimblandy in #2579
- Fixed misspelling in documentation by @zenitopires in #2634
- Update push constant docs to reflect the API by @Noxime in #2637
- Exclude dependencies from documentation by @yutannihilation in #2642
- Document
GpuFuture. by @jimblandy in #2644 - Document random bits and pieces. by @jimblandy in #2651
- Add cross-references to each wgpu type's documentation. by @kpreid in #2653
- RenderPassDescriptor: make label lifetime match doc, and make names descriptive. by @kpreid in #2654
- Document
VertexStepMode. by @jimblandy in #2685 - Add links for SpirV documents. by @huandzh in #2697
- Add symlink LICENSE files into crates. by @dskkato in #2604
- Fix documentation links. by @jimblandy in #2756
- Improve push constant documentation, including internal docs. by @jimblandy in #2764
- Clarify docs for
wgpu_core'sIdandgfx_select!. by @jimblandy in #2766 - Update the Supported Platforms table in README by @jinleili in #2770
- Remove depth image from readme - we don't dictate direction of depth by @cwfitzgerald in #2812
- Update
ashto0.37by @a1phyr in #2557 - Update parking_lot to 0.12. by @emilio in #2639
- Accept both parking-lot 0.11 and 0.12, to avoid windows-rs. by @jimblandy in #2660
- Update web-sys to 0.3.58, sparse attachments support by @jinleili in #2813
- Remove use of inplace_it by @mockersf in #2889
- Clean up features in deno by @crowlKats in #2445
- Dont panic when submitting same commandbuffer multiple times by @crowlKats in #2449
- Handle error sources to display full errors by @crowlKats in #2454
- Pull changes from deno repo by @crowlKats in #2455
- Fix cts_runner by @crowlKats in #2456
- Update deno_webgpu by @crowlKats in #2539
- Custom op arity by @crowlKats in #2542
- Fix conserative-raster low res target getting zero sized on resize by @Wumpf in #2318
- Replace run-wasm-example.sh with aliased rust crate (xtask) by @rukai in #2346
- Get cargo-run-wasm from crates.io by @rukai in #2415
- Fix msaa-line example's unnecessary MSAA data store by @jinleili in #2421
- Make shadow example runnable on iOS Android devices by @jinleili in #2433
- Blit should only draw one triangle by @CurryPseudo in #2474
- Fix wasm examples failing to compile by @Liamolucko in #2524
- Fix incorrect filtering used in mipmap generation by @LaylBongers in #2525
- Correct program output ("Steps", not "Times") by @skierpage in #2535
- Fix resizing behaviour of hello-triangle example by @FrankenApps in #2543
- Switch from
cgmathtoglamin examples by @a1phyr in #2544 - Generate 1x1 mip level by @davidar in #2551
- Wgpu/examples/shadow: Don't run on llvmpipe. by @jimblandy in #2595
- Avoid new WGSL reserved words in wgpu examples. by @jimblandy in #2606
- Move texture-array example over to wgsl by @cwfitzgerald in #2618
- Remove the default features from wgpu-info by @jinleili in #2753
- Fix bunnymark test screenshot and replace rand with nanorand by @stevenhuyn in #2746
- Use FIFO swapchain in examples by @cwfitzgerald in #2790
- Test WebGPU backend with extra features by @kvark in #2362
- Lint deno_webgpu & wgpu-core by @AaronO in #2403
- IdentityManager:
from_indexmethod is unneeded. by @jimblandy in #2424 - Added id32 feature by @caiiiycuk in #2464
- Update dev deps by @rukai in #2493
- Use cargo nextest for running our tests by @cwfitzgerald in #2495
- Many Steps Towards GL Testing Working by @cwfitzgerald in #2504
- Rename ci.txt to ci.yml by @simon446 in #2510
- Re-enable GL testing in CI by @cwfitzgerald in #2508
- Expect shadow example to pass on GL by @kvark in #2541
- Simplify implementation of RefCount and MultiRefCount. by @jimblandy in #2548
- Provide a proper
newmethod forRefCount. by @jimblandy in #2570 - Add logging to LifetimeTracker::triage_suspected. by @jimblandy in #2569
- wgpu-hal: Work around cbindgen bug: ignore
gles::eglmodule. by @jimblandy in #2576 - Specify an exact wasm-bindgen-cli version in publish.yml. by @jimblandy in #2624
- Rename
timeout_ustotimeout_ns, to match actual units. by @jimblandy in #2645 - Move set_index_buffer FFI functions back into wgpu. by @jimblandy in #2661
- New function:
Global::create_buffer_error. by @jimblandy in #2673 - Actually use RenderBundleEncoder::set_bind_group in tests. by @jimblandy in #2678
- Eliminate wgpu_core::commands::bundle::State::raw_dynamic_offsets. by @jimblandy in #2684
- Move RenderBundleEncoder::finish's pipeline layout id into the state. by @jimblandy in #2755
- Expect shader_primitive_index tests to fail on AMD RADV POLARIS12. by @jimblandy in #2754
- Introduce
VertexStep: a stride and a step mode. by @jimblandy in #2768 - Increase max_outliers on wgpu water example reftest. by @jimblandy in #2767
- wgpu_core::command::bundle: Consolidate pipeline and vertex state. by @jimblandy in #2769
- Add type annotation to render pass code, for rust-analyzer. by @jimblandy in #2773
- Expose naga span location helpers by @nical in #2752
- Add create_texture_error by @nical in #2800
- fix crashes when logging in debug message callbacks
- fix program termination when dx12 or gles error messages happen.
- implement validation canary
- DX12:
- Ignore erroneous validation error from DXGI debug layer.
- Metal:
- check for MSL-2.3
- Metal:
- preserve vertex invariance
- Vulkan
- fix stencil read/write masks
- Gles:
- reset index binding properly
- DX12:
- fix copies into 1D textures
- fix tracy compile error
- fix buffer binding limits beyond 2Gb
- fix zero initialization of 3D textures
- Metal:
- fix surface texture views
- Gles:
- extend
libwaylandsearch paths
- extend
- zero initialization uses now render target clears when possible (faster and doesn't enforce COPY_DST internally if not necessary)
- fix use of MSAA targets in WebGL
- fix not providing
COPY_DSTflag for textures causing assertions in some cases - fix surface textures not getting zero initialized
- clear_texture supports now depth/stencil targets
- error message on creating depth/stencil volume texture
- Vulkan:
- fix validation error on debug message types
- DX12:
- fix check for integrated GPUs
- fix stencil subresource transitions
- Metal:
- implement push constants
- API:
MULTIVIEWfeatureDEPTH_CLIP_CONTROLfeature to replace the oldDEPTH_CLAMPTEXTURE_FORMAT_16BIT_NORMfeature- push/pop error scopes on the device
- more limits for compute shaders
SamplerBindingTypeinstead of booleans- sampler arrays are supported by
TEXTURE_BINDING_ARRAYfeature - "glsl" cargo feature for accepting GLSL shader code
- enforced MSRV-1.53
- correctness:
- textures are zero-initialized
- lots and lots of fixes
- validation:
- match texture-sampler pairs
- check
min_binding_sizelate at draw - check formats to match in
copy_texture_to_texture - allow
strip_index_formatto be none if unused - check workgroup sizes and counts
- shaders:
- please refer to naga-0.8 changelog
- nice error messages
- Core:
- validate device descriptor before actually creating it
- fix validation of texture-sampler pairs
- Vulkan:
- fix running on Vulkan-1.1 instance
- improve detection of workaround for Intel+Nvidia on Linux
- fix resource limits on Vulkan-1.2
- fix the check for storage buffer requirement
- change internal semaphore logic to work around Linux+Intel bugs
- fix enabling extension-provided features
- GLES:
- fix running on old and bogus drivers
- fix stale samplers on bindings change
- fix integer textures
- fix querying work group parameters
- fix stale PBO bindings caused by resource copies
- fix rendering to cubemap faces
- fix
Rgba16Floatformat - fix stale vertex attributes when changing the pipeline
- Metal:
- fix window resizing for running in multiple processes
- Web:
- fix
set_index_bufferandset_vertex_bufferto have optional sizes
- fix
- fix buffer transition barriers
- Metal:
- disable RW buffers on macOS 10.11
- fix memory leaks in render pass descriptor
- WebGL:
- fix surface reconfiguration
- GLES:
- fix mapping when persistent mapping isn't supported
- allow presentation in Android emulator
- fix sRGB attributes on EGL-1.4 contexts
- GL:
- fix mapping flags and buffer initialization
- fix context creation when sRGB is available
- fix bind group layout lifetime with regard to bind groups
- GL/WebGL: fix vertex buffer bindings with non-zero first instance
- DX12: fix cube array view construction
- Vulkan: fix NV optimus detection on Linux
- GL:
- fix indirect dispatch buffers
- WebGL:
- fix querying storage-related limits
- work around a browser bug in the clear shader
- Infrastructure:
- Deno WebGPU plugin is a part of the repository
- WebGPU CTS is ran on CI via Deno
- API:
- initial WebGL support
SwapchainFrameis removed.SurfaceTexture::present()needs to be called instead of dropping.- better SPIR-V control flow processing
- ability to request a software (fallback) adapter
- new limits for
min_uniform_buffer_offset_alignmentandmin_storage_buffer_offset_alignment - features:
- new
PARTIALLY_BOUND_BINDING_ARRAY NON_FILL_POLYGON_MODEis split intoPOLYGON_MODE_LINEandPOLYGON_MODE_POINT
- new
- fixes:
- many shader-related fixes in Naga-0.7
- fix a panic in resource cleanup happening when they are dropped on another thread
- Vulkan:
- create SPIR-V per entry point to work around driver bugs
- expose higher descriptor limits based on descriptor indexing capabilities
- GL and Vulkan:
- Fix renderdoc device pointers
- optimization:
- on Vulkan, bounds checks are omitted if the platform can do them natively
- fix
write_texturefor array textures - fix closing an encoder on validation error
- expose Metal surface creation
- panic with an actual error message in the default handler
- Metal:
- fix stencil back-face state
- fix the limit on command buffer count
- Metal:
- fix stencil operations
- fix memory leak on M1 when out of focus
- fix depth clamping checks
- fix unsized storage buffers beyond the first
- Vulkan:
- fix read access barriers for writable storage buffers
- fix shaders using cube array textures
- work around Linux Intel+Nvidia driver conflicts
- work around Adreno bug with
OpName
- DX12:
- fix storage binding offsets
- Metal:
- fix compressed texture copies
- All:
- fix querying the size of storage textures
- Vulkan:
- use render pass labels
- Metal:
- fix moving the surface between displays
- DX12:
- enable BC compressed textures
- GL:
- fix vertex-buffer and storage related limits
- All:
- expose more formats via adapter-specific feature
- fix creation of depth+stencil views
- validate cube textures to not be used as storage
- fix mip level count check for storage textures
- Metal:
- fix usage of work group memory
- DX12:
- critical fix of pipeline layout
- Infrastructure:
gfx-halis replaced by the in-house graphics abstractionwgpu-hal. Backends: Vulkan, Metal, D3D-12, and OpenGL ES-3.- examples are tested automatically for image snapshots.
- API:
crossfeature is removed entirely. Only Rust code from now on.- processing SPIR-V inputs for later translation now requires
spirvcompile feature enabled - new
Features::SPIRV_SHADER_PASSTHROUGHrun-time feature allows providing pass-through SPIR-V (orthogonal to the compile feature) - several bitflag names are renamed to plural:
TextureUsage,BufferUsage,ColorWrite. - the
SwapChainis merged intoSurface. Returned frames areTextureinstead ofTextureView. - renamed
TextureUsagebits:SAMPLED->TEXTURE_BINDING,STORAGE->STORAGE_BINDING. - renamed
InputStepModetoVertexStepMode. - readable storage textures are no longer a part of the base API. Only exposed via format-specific features, non-portably.
- implemented
Rgb9e5Ufloatformat. - added limits for binding sizes, vertex data, per-stage bindings, and others.
- reworked downlevel flags, added downlevel limits.
resolver = "2"is now required in top-level cargo manifests
- Fixed:
Device::create_query_setwould return an error when creating exactlyQUERY_SET_MAX_QUERIES(8192) queries. Now it only returns an error when trying to create more thanQUERY_SET_MAX_QUERIESqueries.
- fix
Features::TEXTURE_SPECIFIC_FORMAT_FEATURESnot being supported for rendertargets
- fix buffer inits delayed by a frame
- fix query resolves to initialize buffers
- fix pipeline statistics stride
- fix the check for maximum query count
- Updated:
- naga to
v0.5.
- naga to
- Added:
Features::VERTEX_WRITABLE_STORAGE.Features::CLEAR_COMMANDSwhich allows you to usecmd_buf.clear_textureandcmd_buf.clear_buffer.
- Changed:
- Updated default storage buffer/image limit to
8from4.
- Updated default storage buffer/image limit to
- Fixed:
Buffer::get_mapped_rangecan now have a range of zero.- Fixed output spirv requiring the "kernel" capability.
- Fixed segfault due to improper drop order.
- Fixed incorrect dynamic stencil reference for Replace ops.
- Fixed tracking of temporary resources.
- Stopped unconditionally adding cubemap flags when the backend doesn't support cubemaps.
- Validation:
- Ensure that if resources are viewed from the vertex stage, they are read only unless
Features::VERTEX_WRITABLE_STORAGEis true. - Ensure storage class (i.e. storage vs uniform) is consistent between the shader and the pipeline layout.
- Error when a color texture is used as a depth/stencil texture.
- Check that pipeline output formats are logical
- Added shader label to log messages if validation fails.
- Ensure that if resources are viewed from the vertex stage, they are read only unless
- Tracing:
- Make renderpasses show up in the trace before they are run.
- Docs:
- Fix typo in
PowerPreference::LowPowerdescription.
- Fix typo in
- Player:
- Automatically start and stop RenderDoc captures.
- Examples:
- Handle winit's unconditional exception.
- Internal:
- Merged wgpu-rs and wgpu back into a single repository.
- The tracker was split into two different stateful/stateless trackers to reduce overhead.
- Added code coverage testing
- CI can now test on lavapipe
- Add missing extern "C" in wgpu-core on
wgpu_render_pass_execute_bundles - Fix incorrect function name
wgpu_render_pass_bundle_indexed_indirecttowgpu_render_bundle_draw_indexed_indirect.
- fix dynamic stencil reference for Replace ops
- fix SPIR-V generation from WGSL, which was broken due to "Kernel" capability
- validate buffer storage classes
- Added support for storage texture arrays for Vulkan and Metal.
- Naga is used by default to translate shaders, SPIRV-Cross is optional behind
crossfeature - Features:
- buffers are zero-initialized
- downlevel limits for DX11/OpenGL support
- conservative rasterization (native-only)
- buffer resource indexing (native-only)
- API adjustments to the spec:
- Renamed
RenderPassColorAttachmentDescriptortoRenderPassColorAttachment:- Renamed the
attachmentmember toview
- Renamed the
- Renamed
RenderPassDepthStencilAttachmentDescriptortoRenderPassDepthStencilAttachment:- Renamed the
attachmentmember toview
- Renamed the
- Renamed
VertexFormatvalues- Examples:
Float3->Float32x3,Ushort2->Uint16x2
- Examples:
- Renamed the
depthvalue ofExtent3dtodepth_or_array_layers - Updated blending options in
ColorTargetState:- Renamed
BlendStatetoBlendComponent - Added
BlendStatestruct to hold color and alpha blend state - Moved
color_blendandalpha_blendmembers intoblendmember
- Renamed
- Moved
clamp_depthfromRastizerStatetoPrimitiveState - Updated
PrimitiveState:- Added
conservativemember for enabling conservative rasterization
- Added
- Updated copy view structs:
- Renamed
TextureCopyViewtoImageCopyTexture - Renamed
TextureDataLayouttoImageDataLayout - Changed
bytes_per_rowandrows_per_imagemembers ofImageDataLayoutfromu32toOption<NonZeroU32>
- Renamed
- Changed
BindingResource::Bindingfrom containing fields directly to containing aBufferBinding - Added
BindingResource::BufferArray
- Renamed
- Infrastructure:
- switch from
tracingtoprofiling - more concrete and detailed errors
- API traces include the command that crashed/panicked
- Vulkan Portability support is removed from Apple platforms
- switch from
- Validation:
- texture bindings
- filtering of textures by samplers
- interpolation qualifiers
- allow vertex components to be underspecified
- expose
wgc::device::queuesub-module in public - fix the indexed buffer check
- fix command allocator race condition
- Major API changes:
RenderPipelineDescriptorBindingType- new
ShaderModuleDescriptor - new
RenderEncoder
- Features:
- (beta) WGSL support, including the ability to bypass SPIR-V entirely
- (beta) implicit bind group layout support
- better error messages
- timestamp and pipeline statistics queries
- ETC2 and ASTC compressed textures
- (beta) targeting Wasm with WebGL backend
- reduced dependencies
- Native-only:
- clamp-to-border addressing
- polygon fill modes
- query a format for extra capabilities
f64support in shaders
- Validation:
- shader interface
- render pipeline descriptor
- vertex buffers
- don't panic in the staging belt if the channel is dropped
- Crates:
- C API is moved to another repository
player: standalone API replayer and tester
- Features:
- Proper error handling with all functions returning
Result - Graceful handling of "error" objects
- API tracing infrastructure
- uploading data with
write_buffer/write_texturequeue operations - reusable render bundles
- read-only depth/stencil attachments
- bind group layout deduplication
- Cows, cows everywhere
- Web+Native features:
- Depth clamping (feature)
- BC texture compression
- Native-only features:
- mappable primary buffers
- texture array bindings
- push constants
- multi-draw indirect
- Proper error handling with all functions returning
- Validation:
- all transfer operations
- all resource creation
- bind group matching to the layout
- experimental shader interface matching with Naga
- add debug markers support
- fix destruction of adapters, swap chains, and bind group layouts
- fix command pool leak with temporary threads
- improve assertion messages
- implement
From<TextureFormat>forTextureComponentType
- fix memory management of staging buffers
- fix reading access to storage textures
- another fix to layout transitions for swapchain images
- fix read-only storage flags
- fix pipeline layout life time
- improve various assert messages
- fix tracking of swapchain images that are used multiple times in a command buffer
- fix tracking of initial usage of a resource across a command buffer
- Crates:
wgpu-types: common types between native and web targetswgpu-core: internal API for the native and remote wrappers
- Features:
- based on gfx-hal-0.5
- moved from Rendy to the new
gfx-memoryandgfx-descriptorcrates - passes are now recorded on the client side. The user is also responsible to keep all resources referenced in the pass up until it ends recording.
- coordinate system is changed to have Y up in the rendering space
- revised GPU lifetime tracking of all resources
- revised usage tracking logic
- all IDs are now non-zero
- Mailbox present mode
- Validation:
- active pipeline
- Fixes:
- lots of small API changes to closely match upstream WebGPU
- true read-only storage bindings
- unmapping dropped buffers
- better error messages on misused swapchain frames
- improved swap chain error handling
- fixed render pass transitions
- fixed depth/stencil transitions
- fixed dynamic offset iteration
- Platforms: removed OpenGL/WebGL support temporarily
- Features:
- based on gfx-hal-0.4 with the new swapchain model
- exposing adapters from all available backends on a system
- tracking of samplers
- cube map support with an example
- Validation:
- buffer and texture usage
- fixed instance creation on Windows
- fixed pipeline barriers that aren't transitions
- Platforms: experimental OpenGL/WebGL
- Crates:
- Rust API is moved out to another repository
- Features:
- based on gfx-hal-0.3 with help of
rendy-memoryandrendy-descriptor - type-system-assisted deadlock prevention (for locking internal structures)
- texture sub-resource tracking
raw-window-handleintegration instead ofwinit- multisampling with an example
- indirect draws and dispatches
- stencil masks and reference values
- native "compute" example
- everything implements
Debug
- based on gfx-hal-0.3 with help of
- Validation
- vertex/index/instance ranges at draw calls
- bing groups vs their expected layouts
- bind group buffer ranges
- required stencil reference, blend color
- fixed frame acquisition GPU waits
- fixed submission tracking
- added support for blend colors
- fixed bind group compatibility at the gfx-hal level
- validating the bind groups and blend colors
- fixed vertex format mapping
- fixed building with "empty" backend on Windows
- bumped the default descriptor pool size
- fixed host mapping alignments
- validating the uniform buffer offset
- Platforms: iOS/Metal, D3D11
- Crates:
wgpu-remote: remoting layer for the cross-process boundarygfx-examples: selected gfx pre-ll examples ported over
- Features:
- native example for compute
- "gfx-cube" and "gfx-shadow" examples
- copies between buffers and textures
- separate object identity for the remote client
- texture view tracking
- native swapchain resize support
- buffer mapping
- object index epochs
- comprehensive list of vertex and texture formats
- validation of pipeline compatibility with the pass
- Fixes
- fixed resource destruction
- Platforms: Linux/Vulkan, Windows/Vulkan, D3D12, macOS/Metal
- Crates:
wgpu-native: C API implementation of WebGPU, based on gfx-halwgpu-bindings: auto-generated C headerswgpu: idiomatic Rust wrapperexamples: native C examples
- Features:
- native examples for triangle rendering
- basic native swapchain integration
- concept of the storage hub
- basic recording of passes and command buffers
- submission-based lifetime tracking and command buffer recycling
- automatic resource transitions