• Advertisement
Sign in to follow this  

Append buffer: buffer byte stride vs resource view byte stride

This topic is 2064 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

Hello all,

I'm getting this error from dx:

"ID3D11DeviceContext::DrawIndexed: The Shader Resource View in slot 0 of the Vertex Shader unit has structure stride 128 while the shader expects a structure stride of 32. This mismatch is invalid if the shader actually uses the view (e.g. it is not skipped due to shader code branching)."

The app works as expected, but I want to avoid the error filling the log. Have done lots of searching on net and experimenting with code and can't get rid of it.

What I've got is an append buffer that the compute shader writes to through a UAV. It's then unbound from compute shader, and bound to the vertex shader as a resource view.

The append buffer is made with a StructureByteStride of 128 bytes. This one struct contains 4 vertices. Need to have 4 vertices in one chunk, so that when vertices are written to the append buffer they are written in one append() as two primitives and aren't mixed up in the append process.

But the vertex shader reads single 32 byte vertices. Here's how the buffer is declared in the vert shader:
StructuredBuffer<cVertIn> vertsIn

So I guess the mismatch is between the 128 byte stride specified when the buffer was created, and the 32 byte stride of the struct declared in the vert shader hlsl.

Is there a way of having different byte strides for a resource view and its associated buffer? Any other way to avoid the error? I have to use an append buffer, because the amount of data the compute shader generates is variable.

Paul

Share this post


Link to post
Share on other sites
Advertisement
If you use a StructuredBuffer then the stride declared in the shader needs to exactly match the stride that you used when you created the actual Buffer resource, you can't create an SRV with a different stride. The only alternatives I know of would be to use a ByteAddressBuffer instead of a StructuredBuffer and manage the append/ingatomic increment yourself, or to copy the buffer contents from one structured buffer to another structured buffer with a different stride.

Share this post


Link to post
Share on other sites
Hi MJP, thanks for your reply. Dash it! I was hoping that there was some way around. I tried copying the buffer to another some weeks ago, but that was just too slow (the buffer is very large). I might give the raw buffer + atomic increment thing a go. Or change the vertex buffer to accept a 128 byte struct, and read from within that struct the one vert it needs. Thanks for the advice.

Share this post


Link to post
Share on other sites
If you only read from a portion of your structure, then the generated assembly will only have enough memory loads to cover that portion of the structure. So in your case if your vertex shader only uses 32 bytes, the assembly will only load 32 bytes.

Share this post


Link to post
Share on other sites
Meaning that then the vertex shader will still report to dx a 32 byte stride? I assumed that the hlsl compiler looked at which struct was used with the StructuredBuffer. Oh well. Perhaps I'll live with the warning, or work out how to use a raw buffer with a counter. Thanks again!

Share this post


Link to post
Share on other sites
No what I mean is that the shader won't fetch all 128 bytes of the structure if your shader doesn't use all of it, even if the stride in HLSL is 128 bytes. You'd still need to calculate the buffer index based on the stride of 128 bytes, but you can do something like this:


struct Vertex
{
float3 Position;
float3 Normal;
...
};


struct VertexGroup
{
Vertex Vertices[4];
};

StructuredBuffer Vertices<VertexGroup>;

void VSMain(in uint VertexID : SV_VertexID)
{
Vertex vtx = Vertices[VertexID / 4].Vertices[VertexID % 4];
...
}


Then it that case the vertex shader will only access 1 vertex worth of data. Edited by MJP

Share this post


Link to post
Share on other sites
Ah, thanks, that's how I thought I'd have to access the object.

Share this post


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

  • Advertisement