Alright, I'm getting back into trying to run a shader through DirectX. This time I'm looking at DirectX11 and DirectCompute as it was suggested. It appears to make it simpler than with DX9 to run the shader without needing to configure tons of "stuff" that isn't directly related to what I'm doing (like vertexes).

Now I have a question as to how to configure the shader parameters. In DirectX 9, this works

m_pPixelConstantTable->SetInt(m_pDevice, name, value);
m_pPixelConstantTable->SetFloat(m_pDevice, name, value);
m_pPixelConstantTable->SetBool(m_pDevice, name, value);

In DirectX 11, it appers I have to instead use ID3D11DeviceContext::PSSetConstantBuffers and create a ID3D11Buffer for each parameter. This would require considerably more code, and more importantly... how does the shader know the data type of a parameter?

What's the simplest syntax for defining parameters for the shader?

Thanks

In DirectX 11, it appers I have to instead use ID3D11DeviceContext::PSSetConstantBuffers and create a ID3D11Buffer for each parameter. This would require considerably more code, and more importantly... how does the shader know the data type of a parameter?

No, no.

You create a constant buffer in shader with all your variables in it like this

cbuffer cbPerFrameData : register(b0)

{

float3 SomeData;

float3 AnotherData;

};

and you create same struct in your C++ code like this:

struct CBPerFrameData

{

XMFLOAT4 SomeData;

XMFLOAT4 AnotherData;

};

and you create constant buffer of it using the bind flag D3D11_BIND_CONSTANT_BUFFER and then you use PSSetConstantBuffers or VSSetConstantBuffers to upload it to the shader.

Check this out:

https://github.com/newtechnology/NT-Engine/blob/master/NT%20Engine/ConstantBuffer.h

https://github.com/newtechnology/NT-Engine/blob/master/NT%20Engine/ConstantBuffer.cpp

https://github.com/newtechnology/NT-Engine/blob/master/NT%20Engine/Shaders.h

https://github.com/newtechnology/NT-Engine/tree/master/NT%20Engine/Shaders

I don't get to define how they are being received in the HLSL shaders as I'm using pre-written shaders

https://github.com/zachsaw/MPDN_Extensions/blob/master/Extensions/RenderScripts/SuperRes/SuperResEx.hlsl

sampler s0    : register(s0);
sampler sDiff : register(s1);
float4 p0      : register(c0);
float2 p1      : register(c1);
float4 size1  : register(c2); // Original size
float4 args0  : register(c3);

In this case, do I have to create a buffer for each one? Also it is using PS_3_0

The link you posted is not valid, but I suppose that I have to use PSSetShaderResources for samplers?

https://msdn.microsoft.com/en-us/library/windows/desktop/ff476473%28v=vs.85%29.aspx

So basically, I only need to create one buffer, with the byte order being the same as in the shader code?

2 questions.

1. How do I define the buffer as being the default named "$Global" ?

2. In the sample I posted, the data types are float4, float2, float4 and float4. How do I use these data types in C++? Is it full float or half-float?

And in my case, parameters are assigned dynamically to run any shader (at least those who follow this parameters format) so I can't hard-code it as a struct. I suppose it will work just the same by filling the buffer with the right data in the right order?

How do I set the buffer variables without setting a buffer constant? Should I use PSGetConstantBuffers(0, 1) instead of PSSetConstantBuffers to get a buffer already created by default?

DirectCompute was looking easy so far, but now configuring its parameters seems to be a real headache.

Creating and passing a texture object is pretty easy, but if I have to instead use PSSetSamplers, how do I bind that to the texture? I'm not finding any argument for that in CreateSamplerState or D3D11_SAMPLER_DESC. It seems I simply have to call CreateSamplerState instead of CreateShaderResourceView after creating the texture, except that I don't see where to pass in the texture.

And then, will DX11 fully support running PS_3_0 shaders? It seems to be designed for PS_4_0.