Add half-precision matrix type aliases in GLSL (#7066)

Fixes #6708

This commit adds type aliases for half-precision matrices, including
f16mat3x2, f16mat3x3, f16mat3x4, f16mat4x2, f16mat4x3, and f16mat4x4.
Convenience aliases for square matrices (f16mat2, f16mat3, f16mat4) are
also added.

This commit introduces a new test file that validates the usage of
half-precision types in a compute shader context.

1 files changed, 12 insertions, 0 deletions

diff --git a/source/slang/glsl.meta.slang b/source/slang/glsl.meta.slang
index 05cef836c..2d9078855 100644
--- a/source/slang/glsl.meta.slang
+++ b/source/slang/glsl.meta.slang
@@ -101,6 +101,18 @@ public typealias f16mat2x2 = matrix<half, 2, 2>;
 public typealias f16mat2x3 = matrix<half, 2, 3>;
 public typealias f16mat2x4 = matrix<half, 2, 4>;
 
+public typealias f16mat3x2 = matrix<half, 3, 2>;
+public typealias f16mat3x3 = matrix<half, 3, 3>;
+public typealias f16mat3x4 = matrix<half, 3, 4>;
+
+public typealias f16mat4x2 = matrix<half, 4, 2>;
+public typealias f16mat4x3 = matrix<half, 4, 3>;
+public typealias f16mat4x4 = matrix<half, 4, 4>;
+
+// Convenience aliases for square matrices
+public typealias f16mat2 = f16mat2x2;
+public typealias f16mat3 = f16mat3x3;
+public typealias f16mat4 = f16mat4x4;
 
 public out float4 gl_Position : SV_Position;
 public out float gl_PointSize : SV_PointSize;