Matrix4x3: add Matrix4x3, needed for raytracing intrinsics

Author: Firestar99

crates/spirv-std/src/lib.rs

@@ -95,6 +95,7 @@ pub mod float;
 pub mod image;
 pub mod indirect_command;
 pub mod integer;
+pub mod matrix;
 pub mod memory;
 pub mod number;
 pub mod ray_tracing;

crates/spirv-std/src/matrix.rs

@@ -0,0 +1,78 @@
+//! a set of common SPIR-V Matrices, used for intrinsics
+
+use glam::{Affine3A, Mat3, Mat3A, Mat4, Vec3, Vec3A};
+
+/// A Matrix with 4 columns of [`Vec3`], very similar to glam's [`Affine3A`].
+///
+/// Primarily used in ray tracing extensions to represent object rotation, scale and translation.
+#[derive(Clone, Copy, Default, PartialEq)]
+#[repr(C)]
+#[spirv(matrix)]
+#[allow(missing_docs)]
+pub struct Matrix4x3 {
+    pub x: Vec3A,
+    pub y: Vec3A,
+    pub z: Vec3A,
+    pub w: Vec3A,
+}
+
+/// The `from_*` fn signatures should match [`Affine3A`], to make it easier to switch to [`Affine3A`] later.
+/// The `to_*` fn signatures are custom
+impl Matrix4x3 {
+    /// Convert from glam's [`Affine3A`]
+    pub fn from_affine3a(affine: Affine3A) -> Self {
+        Self {
+            x: affine.x_axis,
+            y: affine.y_axis,
+            z: affine.z_axis,
+            w: affine.w_axis,
+        }
+    }
+
+    /// Creates an affine transform from a 3x3 matrix (expressing scale, shear and
+    /// rotation)
+    pub fn from_mat3(mat3: Mat3) -> Self {
+        Self::from_affine3a(Affine3A::from_mat3(mat3))
+    }
+
+    /// Creates an affine transform from a 3x3 matrix (expressing scale, shear and rotation)
+    /// and a translation vector.
+    ///
+    /// Equivalent to `Affine3A::from_translation(translation) * Affine3A::from_mat3(mat3)`
+    pub fn from_mat3_translation(mat3: Mat3, translation: Vec3) -> Self {
+        Self::from_affine3a(Affine3A::from_mat3_translation(mat3, translation))
+    }
+
+    /// The given `Mat4` must be an affine transform,
+    /// i.e. contain no perspective transform.
+    pub fn from_mat4(m: Mat4) -> Self {
+        Self::from_affine3a(Affine3A::from_mat4(m))
+    }
+
+    /// Convert to glam's [`Affine3A`]
+    pub fn to_affine3a(self) -> Affine3A {
+        Affine3A {
+            matrix3: Mat3A {
+                x_axis: self.x,
+                y_axis: self.y,
+                z_axis: self.z,
+            },
+            translation: self.w,
+        }
+    }
+
+    /// Creates a 3x3 matrix representing the rotation and scale, cutting off the translation
+    pub fn to_mat3a(self) -> Mat3A {
+        self.to_affine3a().matrix3
+    }
+
+    /// Creates a 3x3 matrix representing the rotation and scale, cutting off the translation
+    pub fn to_mat3(self) -> Mat3 {
+        Mat3::from(self.to_mat3a())
+    }
+
+    /// Creates a 4x4 matrix from this affine transform
+    pub fn to_mat4(self) -> Mat4 {
+        Mat4::from(self.to_affine3a())
+    }
+}

crates/spirv-std/src/ray_tracing.rs

@@ -3,6 +3,7 @@
 // NOTE(eddyb) "&-masking with zero", likely due to `NONE = 0` in `bitflags!`.
 #![allow(clippy::bad_bit_mask)]
 
+use crate::matrix::Matrix4x3;
 use crate::vector::Vector;
 #[cfg(target_arch = "spirv")]
 use core::arch::asm;
@@ -1002,22 +1003,14 @@ impl RayQuery {
     #[spirv_std_macros::gpu_only]
     #[doc(alias = "OpRayQueryGetIntersectionObjectToWorldKHR")]
     #[inline]
-    pub unsafe fn get_candidate_intersection_object_to_world<V: Vector<f32, 3>>(&self) -> [V; 4] {
+    pub unsafe fn get_candidate_intersection_object_to_world(&self) -> Matrix4x3 {
         unsafe {
             let mut result = Default::default();
 
             asm! {
                 "%u32 = OpTypeInt 32 0",
-                "%f32 = OpTypeFloat 32",
-                "%f32x3 = OpTypeVector %f32 3",
-                "%f32x3x4 = OpTypeMatrix %f32x3 4",
                 "%intersection = OpConstant %u32 0",
-                "%matrix = OpRayQueryGetIntersectionObjectToWorldKHR %f32x3x4 {ray_query} %intersection",
-                "%col0 = OpCompositeExtract %f32x3 %matrix 0",
-                "%col1 = OpCompositeExtract %f32x3 %matrix 1",
-                "%col2 = OpCompositeExtract %f32x3 %matrix 2",
-                "%col3 = OpCompositeExtract %f32x3 %matrix 3",
-                "%result = OpCompositeConstruct typeof*{result} %col0 %col1 %col2 %col3",
+                "%result = OpRayQueryGetIntersectionObjectToWorldKHR typeof*{result} {ray_query} %intersection",
                 "OpStore {result} %result",
                 ray_query = in(reg) self,
                 result = in(reg) &mut result,
@@ -1037,22 +1030,14 @@ impl RayQuery {
     #[spirv_std_macros::gpu_only]
     #[doc(alias = "OpRayQueryGetIntersectionObjectToWorldKHR")]
     #[inline]
-    pub unsafe fn get_committed_intersection_object_to_world<V: Vector<f32, 3>>(&self) -> [V; 4] {
+    pub unsafe fn get_committed_intersection_object_to_world(&self) -> Matrix4x3 {
         unsafe {
             let mut result = Default::default();
 
             asm! {
                 "%u32 = OpTypeInt 32 0",
-                "%f32 = OpTypeFloat 32",
-                "%f32x3 = OpTypeVector %f32 3",
-                "%f32x3x4 = OpTypeMatrix %f32x3 4",
                 "%intersection = OpConstant %u32 1",
-                "%matrix = OpRayQueryGetIntersectionObjectToWorldKHR %f32x3x4 {ray_query} %intersection",
-                "%col0 = OpCompositeExtract %f32x3 %matrix 0",
-                "%col1 = OpCompositeExtract %f32x3 %matrix 1",
-                "%col2 = OpCompositeExtract %f32x3 %matrix 2",
-                "%col3 = OpCompositeExtract %f32x3 %matrix 3",
-                "%result = OpCompositeConstruct typeof*{result} %col0 %col1 %col2 %col3",
+                "%result = OpRayQueryGetIntersectionObjectToWorldKHR typeof*{result} {ray_query} %intersection",
                 "OpStore {result} %result",
                 ray_query = in(reg) self,
                 result = in(reg) &mut result,

tests/compiletests/ui/arch/debug_printf_type_checking.stderr

@@ -75,9 +75,9 @@ help: the return type of this call is `u32` due to the type of the argument pass
     |                             |
     |                             this argument influences the return type of `debug_printf_assert_is_type`
 note: function defined here
-   --> $SPIRV_STD_SRC/lib.rs:133:8
+   --> $SPIRV_STD_SRC/lib.rs:134:8
     |
-133 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
+134 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
     |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
     = note: this error originates in the macro `debug_printf` (in Nightly builds, run with -Z macro-backtrace for more info)
 help: change the type of the numeric literal from `u32` to `f32`
@@ -103,9 +103,9 @@ help: the return type of this call is `f32` due to the type of the argument pass
     |                             |
     |                             this argument influences the return type of `debug_printf_assert_is_type`
 note: function defined here
-   --> $SPIRV_STD_SRC/lib.rs:133:8
+   --> $SPIRV_STD_SRC/lib.rs:134:8
     |
-133 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
+134 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
     |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
     = note: this error originates in the macro `debug_printf` (in Nightly builds, run with -Z macro-backtrace for more info)
 help: change the type of the numeric literal from `f32` to `u32`
@@ -131,12 +131,12 @@ error[E0277]: the trait bound `{float}: Vector<f32, 2>` is not satisfied
               `UVec3` implements `Vector<u32, 3>`
             and 5 others
 note: required by a bound in `debug_printf_assert_is_vector`
-   --> $SPIRV_STD_SRC/lib.rs:140:8
+   --> $SPIRV_STD_SRC/lib.rs:141:8
     |
-138 | pub fn debug_printf_assert_is_vector<
+139 | pub fn debug_printf_assert_is_vector<
     |        ----------------------------- required by a bound in this function
-139 |     TY: crate::scalar::Scalar,
-140 |     V: crate::vector::Vector<TY, SIZE>,
+140 |     TY: crate::scalar::Scalar,
+141 |     V: crate::vector::Vector<TY, SIZE>,
     |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ required by this bound in `debug_printf_assert_is_vector`
     = note: this error originates in the macro `debug_printf` (in Nightly builds, run with -Z macro-backtrace for more info)
 
@@ -157,9 +157,9 @@ help: the return type of this call is `Vec2` due to the type of the argument pas
     |                             |
     |                             this argument influences the return type of `debug_printf_assert_is_type`
 note: function defined here
-   --> $SPIRV_STD_SRC/lib.rs:133:8
+   --> $SPIRV_STD_SRC/lib.rs:134:8
     |
-133 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
+134 | pub fn debug_printf_assert_is_type<T>(ty: T) -> T {
     |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
     = note: this error originates in the macro `debug_printf` (in Nightly builds, run with -Z macro-backtrace for more info)
 

tests/compiletests/ui/arch/ray_query_get_intersection_object_to_world_khr.rs

@@ -2,6 +2,7 @@
 // compile-flags: -Ctarget-feature=+RayQueryKHR,+ext:SPV_KHR_ray_query
 
 use glam::Vec3;
+use spirv_std::matrix::Matrix4x3;
 use spirv_std::ray_tracing::{AccelerationStructure, RayFlags, RayQuery};
 use spirv_std::spirv;
 
@@ -10,7 +11,7 @@ pub fn main(#[spirv(descriptor_set = 0, binding = 0)] accel: &AccelerationStruct
     unsafe {
         spirv_std::ray_query!(let mut handle);
         handle.initialize(accel, RayFlags::NONE, 0, Vec3::ZERO, 0.0, Vec3::ZERO, 0.0);
-        let matrix: [glam::Vec3; 4] = handle.get_candidate_intersection_object_to_world();
-        let matrix: [glam::Vec3; 4] = handle.get_committed_intersection_object_to_world();
+        let matrix: Matrix4x3 = handle.get_candidate_intersection_object_to_world();
+        let matrix: Matrix4x3 = handle.get_committed_intersection_object_to_world();
     }
 }

tests/compiletests/ui/spirv-attr/all-builtins.rs

@@ -3,17 +3,9 @@
 // compile-flags: -Ctarget-feature=+DeviceGroup,+DrawParameters,+FragmentBarycentricNV,+FragmentBarycentricKHR,+FragmentDensityEXT,+FragmentFullyCoveredEXT,+Geometry,+GroupNonUniform,+GroupNonUniformBallot,+MeshShadingNV,+MultiView,+MultiViewport,+RayTracingKHR,+SampleRateShading,+ShaderSMBuiltinsNV,+ShaderStereoViewNV,+StencilExportEXT,+Tessellation,+ext:SPV_AMD_shader_explicit_vertex_parameter,+ext:SPV_EXT_fragment_fully_covered,+ext:SPV_EXT_fragment_invocation_density,+ext:SPV_EXT_shader_stencil_export,+ext:SPV_KHR_ray_tracing,+ext:SPV_NV_fragment_shader_barycentric,+ext:SPV_NV_mesh_shader,+ext:SPV_NV_shader_sm_builtins,+ext:SPV_NV_stereo_view_rendering
 
 use spirv_std::glam::*;
+use spirv_std::matrix::Matrix4x3;
 use spirv_std::spirv;
 
-#[derive(Clone, Copy)]
-#[spirv(matrix)]
-pub struct Matrix4x3 {
-    pub x: glam::Vec3,
-    pub y: glam::Vec3,
-    pub z: glam::Vec3,
-    pub w: glam::Vec3,
-}
-
 #[spirv(tessellation_control)]
 pub fn tessellation_control(
     #[spirv(invocation_id)] invocation_id: u32,