RWByteAddressBuffer.LoadAligned

Description

Load a value with type T from the buffer at the specified location with a known alignment.

Signature

T RWByteAddressBuffer.LoadAligned<T>(
    uint location,
    uint alignment);

T RWByteAddressBuffer.LoadAligned<T>(uint location);

Generic Parameters

T

The type of the value to load from the buffer.

Parameters

location : uint

The input address in bytes, which must be a multiple of alignment.

alignment : uint

The known alignment of location in bytes; a compile-time constant.

Return value

The value loaded from the buffer.

Remarks

alignment selects the widest power-of-two-width access it permits: a single wide load when it covers the whole T, otherwise the widest aligned chunks with per-component loads for any remainder. The rule applies to each vector within T: a struct is loaded field by field, a 3-component vector is always split because no power-of-two alignment is a multiple of its size, and a partially covered bool vector is loaded per component. For example, LoadAligned<float4>(location, 8) lowers as follows:

• HLSL with a Shader Model 5.0 or earlier profile: four uint loads, each converted with asfloat.
• HLSL with a Shader Model 5.1 or later profile, SPIR-V, GLSL, CUDA and C++: two float2 loads.
• Metal, WGSL and LLVM-based CPU targets: four per-component loads, for any valid alignment.

Availability and Requirements

Defined for the following targets:

hlsl

Available in all stages.

glsl

Available in all stages.

cpp

Available in all stages.

cuda

Available in all stages.

metal

Available in all stages.

wgsl

Available in all stages.

spirv

Available in all stages.

llvm

Available in all stages.