I have a problem where my IA's output is not the same as my VS's input.

My VS is very straightforward, taking in a position, and outputting a position:

float4 main(float3 position : POSITION, float2 texcoord : TEXCOORD, float3 normal : NORMAL, float3 tangent : TANGENT) : SV_POSITION {
	return float4(position, 1.0f);
}

I thought that it could be the format in my input layout, but I do not see a problem with it.

static constexpr D3D12_INPUT_ELEMENT_DESC input_layout[] = {
 	{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0,        D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0 },
 	{ "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT,  0, D3D12_APPEND_ALIGNED_ELEMENT, D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0 },
 	{ "NORMAL",  0, DXGI_FORMAT_R32G32B32_FLOAT, 0, D3D12_APPEND_ALIGNED_ELEMENT, D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0 },
 	{ "TANGENT", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, D3D12_APPEND_ALIGNED_ELEMENT, D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0 }
};

My VS file is “vs_5_0”, just like how I compile it.

I do not see really any other option that could be affecting the VS and its inputs.

Remember that if no optional stages are enabled before the rasterizer, the vertex shader must output vertices in homogeneous clip coordinates.

How is the cube defined in the vertex buffer?