Jump to content
  • Advertisement
Sign in to follow this  
DustinB

DX12 [D3D12] Problems loading pre-compiled shaders

This topic is 832 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I am attempting to load pre-compiled shaders and create a PSO.  The moment I go to create the PSO, I receive the following D3D12 Error:

D3D12 ERROR: ID3D12Device::CreateInputLayout: Encoded Signature size doesn't match specified size. [ STATE_CREATION ERROR #63: CREATEINPUTLAYOUT_UNPARSEABLEINPUTSIGNATURE].

 

I have the following shaders which compile without errors in VS using Shader Model 5.1:

//VertexShader.hlsl

#include "PSInput.hlsli"

PSInput VSMain(float4 position : POSITION, float4 color : COLOR)
{
	PSInput result;

	result.position = position;
	result.color = color;

	return result;
}

//-------------------------------------------------
// PixelShader.hlsl

#include "PSInput.hlsli"

float4 PSMain(PSInput input) : SV_TARGET
{
	return input.color;
}

//-------------------------------------------------
// PSInput.hlsli

struct PSInput {
    float4 position : SV_POSITION;
    float4 color : COLOR;
};


.

Below is my code for reading in each shader .cso file and binding the shader bytecode to the PSO descriptor. I've checked that both vertexShaderData and pixelShaderData are non-null and that the data length is greater than zero.  Interestingly vertexShaderDataLength=668 bytes, while pxielShaderDataLength=14368 bytes (much greater than I expected so I wonder if this is something to worry about).

byte * vertexShaderData(nullptr);
uint vertexShaderDataLength(0);
ThrowIfFailed(
    ReadDataFromFile(
        GetAssetFullPath(L"VertexShader.cso").c_str(),
        &vertexShaderData,
        &vertexShaderDataLength
    )
 );


byte * pixelShaderData(nullptr);
uint pixelShaderDataLength(0);
ThrowIfFailed(
    ReadDataFromFile(
        GetAssetFullPath(L"PixelShader.cso").c_str(),
        &pixelShaderData,
        &pixelShaderDataLength
    )
 );

    
D3D12_GRAPHICS_PIPELINE_STATE_DESC psoDesc = {};
psoDesc.InputLayout = { inputElementDescriptor, _countof(inputElementDescriptor) };
psoDesc.pRootSignature = rootSignature;
psoDesc.VS = CD3DX12_SHADER_BYTECODE(vertexShaderData, vertexShaderDataLength);
psoDesc.PS = CD3DX12_SHADER_BYTECODE(pixelShaderData, pixelShaderDataLength);
// fill in remainder psoDesc ... 


device->CreateGraphicsPipelineState (     // <--- Exception Thrown here.
    &psoDesc,
    IID_PPV_ARGS(&pipelineState)
);

.

Since the report error mentioned unparsable input signature, I'll include my input element descriptor as well.  I'm using vertex interleaving with just positions and colors.

// Define the vertex input layout.
D3D12_INPUT_ELEMENT_DESC inputElementDescriptor[2];

// Positions
inputElementDescriptor[0].SemanticName = "POSITION";
inputElementDescriptor[0].SemanticIndex = 0;
inputElementDescriptor[0].Format = DXGI_FORMAT_R32G32B32_FLOAT;
inputElementDescriptor[0].InputSlot = 0;
inputElementDescriptor[0].AlignedByteOffset = 0;
inputElementDescriptor[0].InputSlotClass = D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA;
inputElementDescriptor[0].InstanceDataStepRate = 0;

// Colors
inputElementDescriptor[1].SemanticName = "COLOR";
inputElementDescriptor[1].SemanticIndex = 0;
inputElementDescriptor[1].Format = DXGI_FORMAT_R32G32B32_FLOAT;
inputElementDescriptor[1].InputSlot = 0;
inputElementDescriptor[1].AlignedByteOffset = sizeof(float) * 3;
inputElementDescriptor[1].InputSlotClass = D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA;
inputElementDescriptor[1].InstanceDataStepRate = 0;

.

The really odd part of this is that my program runs without error when I combine both the VertexShader and PixelShader into a single effects file "shader.hlsl" and compile the shaders at runtime.

ComPtr<ID3DBlob> vertexShader;
ComPtr<ID3DBlob> pixelShader;


// Compile vertex shader
ThrowIfFailed(
    D3DCompileFromFile (
        GetAssetFullPath(L"shaders.hlsl").c_str(),
        nullptr, nullptr, "VSMain", "vs_5_0", 
        compileFlags, 0, &vertexShader, nullptr
    )
);

// Compile pixel shader
ThrowIfFailed (
    D3DCompileFromFile (
        GetAssetFullPath(L"shaders.hlsl").c_str(),
        nullptr, nullptr, "PSMain", "ps_5_0",
        compileFlags, 0, &pixelShader, nullptr
    )
);

Edited by DustinB

Share this post


Link to post
Share on other sites
Advertisement

I tracked down this piece of information stating HLSL changes to Shader Model 5.1 requiring a specified root signature in order to compile shaders offline:

Note  For shader model 5.1 a root signature must be specified in order to compile shaders offline.

https://msdn.microsoft.com/en-us/library/windows/desktop/dn933268(v=vs.85).aspx

 

Specifying Root Signature in HLSL:

https://msdn.microsoft.com/en-us/library/windows/desktop/dn913202(v=vs.85).aspx

 

I'll give this a try and see if it fixes my issues.

Edited by DustinB

Share this post


Link to post
Share on other sites

You don't need to specify a root signature to compile sm 5.1 shaders. I'm pretty sure that bit in the documentation is a mix-up, since it sounds like it comes from the XB1 SDK. I suspect that something is wrong with how you pre-compile the shaders, or with how you load the data.

 

Share this post


Link to post
Share on other sites

Look at your vertex shader params and the input layout: 

//VertexShader.hlsl

#include "PSInput.hlsli"

PSInput VSMain(float4 position : POSITION, float4 color : COLOR)
{
	PSInput result;

	result.position = position;
	result.color = color;

	return result;
}

Since the report error mentioned unparsable input signature, I'll include my input element descriptor as well.  I'm using vertex interleaving with just positions and colors.

// Define the vertex input layout.
D3D12_INPUT_ELEMENT_DESC inputElementDescriptor[2];

// Positions
inputElementDescriptor[0].SemanticName = "POSITION";
inputElementDescriptor[0].SemanticIndex = 0;
inputElementDescriptor[0].Format = DXGI_FORMAT_R32G32B32_FLOAT;
inputElementDescriptor[0].InputSlot = 0;
inputElementDescriptor[0].AlignedByteOffset = 0;
inputElementDescriptor[0].InputSlotClass = D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA;
inputElementDescriptor[0].InstanceDataStepRate = 0;

// Colors
inputElementDescriptor[1].SemanticName = "COLOR";
inputElementDescriptor[1].SemanticIndex = 0;
inputElementDescriptor[1].Format = DXGI_FORMAT_R32G32B32_FLOAT;
inputElementDescriptor[1].InputSlot = 0;
inputElementDescriptor[1].AlignedByteOffset = sizeof(float) * 3;
inputElementDescriptor[1].InputSlotClass = D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA;
inputElementDescriptor[1].InstanceDataStepRate = 0;

 

You are using float4 for position and color in the vertex shader, but the inputlayout says it's float3.

Share this post


Link to post
Share on other sites

I've filed a bug and asked that this yellow box be removed from the documentation; it does indeed refer to the Xbox One XDK.

 

Using float4 in the shader is fine, it'll simply insert 1.0f into the 'w' channel.

Share this post


Link to post
Share on other sites

If I compile those VS and PS shaders offline to an object file I get sizes of 608 and 532 bytes respectively. I'm not sure how you managed to get a 14KB pixel shader object from just "return input.color". Are you sure the size on disk for PixelShader.cso is ~14KB?

Share this post


Link to post
Share on other sites

You are using float4 for position and color in the vertex shader, but the inputlayout says it's float3.

 

Which is perfectly fine and results in the shader receiving float4(x, y, z, 1).

Share this post


Link to post
Share on other sites

If I compile those VS and PS shaders offline to an object file I get sizes of 608 and 532 bytes respectively. I'm not sure how you managed to get a 14KB pixel shader object from just "return input.color". Are you sure the size on disk for PixelShader.cso is ~14KB?

 

It seems the bloat in size of the PS was due to setting the "enable debugging information" (/Zi) compiler flag.  I turned this off and the PixelShader.cso file compiled to only 532 bytes, matching your size output (I verified this from the filesystem as well).  I also tried disabling debugging information for my VS and VertexShader.cso is only 176 bytes.  With debugging information turned off I am still getting the D3D12 error.

 

As MJP mentioned, this definitely seems to be an issue only with compiling offline.  If I compile the shaders at runtime there are no D3D12 errors.

Edited by DustinB

Share this post


Link to post
Share on other sites

I've filed a bug and asked that this yellow box be removed from the documentation; it does indeed refer to the Xbox One XDK.

 

Using float4 in the shader is fine, it'll simply insert 1.0f into the 'w' channel.

 

Adam, is there an official link for filing bugs?  It would be nice to see if any DX team members could chime in. 

Edited by DustinB

Share this post


Link to post
Share on other sites

Adam, is there an official link for filing bugs?  It would be nice to see if any DX team members could chime in. 

 

At the bottom of every MSDN page there's a "Was this page helpful?", click No and explain in the Additional Feedback box what's up and we'll get it fixed.

 

I've had a look at the code you've posted and nothing obvious stands out, but nothing beats a standalone, single-file repro of the issue and then I/we can run it for ourselves and find out what's up.

Edited by Adam Miles

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!