1
\$\begingroup\$

I have an effect code with a basic structure like

technique TechniqueName {
    pass FirstPass {
        Profile = fx_4_0
        VertexShader = RenderFirstVS
        GeometryShader = null
        PixelShader = RenderFirstPS
    }
    pass SecondPass {
        Profile = fx_4_0
        VetrexShader = RenderSecondVS
        GeometryShader = null
        PixelShader = RenderSecondPS
    }
    pass ThirdPass {
        Profile = fx_4_0
        VertexShader = RenderThirdVS
        GeometryShader = null
        PixelShader = RenderThirdPS
    }
}

Now I tried to compile this with

using (BinaryReader reader = new BinaryReader(stream)) {
    CompilationResult result = ShaderBytecode.Compile(reader.ReadBytes((int)stream.Length), "fx_4_0");
    if (result.HasErrors) {
        throw new Exception(result.Message, new Exception(result.ResultCode.ToString()));
    }
    Data = result.Bytecode.Data;
}

stream is new MemoryStream(Encoding.Default.GetBytes(effectContent)).
The Data-Property (byte[]) is about 1 KiB large but if I try to load it via

context.InputAssembler.InputLayout = new InputLayout(device, effect.Data, someElements)

it crashes with following exception D3D11 ERROR: ID3D11Device::CreateInputLayout: Input Signature in bytecode could not be parsed. Data may be corrupt or in an unrecognizable format. [ STATE_CREATION ERROR #161: CREATEINPUTLAYOUT_UNPARSEABLEINPUTSIGNATURE]

Any idea how I can fix this or why the error is thrown? I do not want to use multiple shaders because I reuse many parameters I do not want to reassign in a deferred shading setup.

\$\endgroup\$

1 Answer 1

Reset to default
1
\$\begingroup\$

When you use the effect shader syntax (techniques and passes), the resulting bytecode is actually a container that includes the bytecode for each compiled vertex and pixel shader for each technique and pass. When creating an input layout, you need to use the bytecode for an individual vertex shader. You'll have to extract that from the effect shader.

It looks like in SharpDX you do this by creating an Effect object using the effect bytecode. You should be able to iterate through the techniques and passes in the effect, although it's not clear to me from the docs just how that works. If you get the EffectPassDescription for a pass, that has a Signature element that should be what you can use to create the input layout.

\$\endgroup\$
4
  • \$\begingroup\$ You referenced for Direct3D10 which is not applicable for the application (I'm using D3D11). So I checked out the D3D11 doc and had to realize it is not accepting the shaderbytecode (throwing SharpDX.SharpDXException with "D3D11CreateEffectFromFile"). \$\endgroup\$ Commented Dec 12, 2013 at 19:38
  • \$\begingroup\$ @AliveDevil So the bytecode that comes out of compiling with fx_4_0 is not accepted when you try to create an Effect object? What if you use fx_5_0 instead? That's the D3D11 version of the effect framework (fx_4_0 is for D3D10). \$\endgroup\$ Commented Dec 12, 2013 at 20:32
  • \$\begingroup\$ I tried it with fx_5_0 but it does not change anything (Data is ~15 Bytes smaller than fx_4_0). \$\endgroup\$ Commented Dec 12, 2013 at 21:30
  • \$\begingroup\$ @AliveDevil Well, that's a new one on me. Perhaps something weird is going on in SharpDX to corrupt the bytecode. I've only done this with stock D3D11, in C++. \$\endgroup\$ Commented Dec 13, 2013 at 2:02

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.