• FEATURED

View more

View more

View more

### Image of the Day Submit

IOTD | Top Screenshots

### The latest, straight to your Inbox.

Subscribe to GameDev.net Direct to receive the latest updates and exclusive content.

# ID3dX11Effect constant buffer

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

3 replies to this topic

### #1osiris11  Members

Posted 20 March 2012 - 04:31 AM

Hey.

I am investigating the use of constant buffers in effect files. Say, my effect file had buffer declarations like this:

cbuffer cbPerViewBuffer
{
matrix viewMatrix;
};


When initializing my shader, i would get a pointer to this effect variable by name, and store it in my object:


ID3DX11EffectMatrixVariable *p_view;
ID3DX11Effect* p_effect;

//... Compiling effect ...//

p_view = p_effect->GetVariableByName("viewMatrix")->AsMatrix();


On rendering, I do

D3DXMATRIX v;
//... doing the math on v ...//

p_view->SetMatrix( v);


There is no buffer mapping or else involved. So my question is, is this even using the advantages of constant buffers, or is this an absurdely abuse of mentioned features?
Should I instead create a custom buffer type, create a buffer of this type, bind it as dynamic to the effect, and use map on each render call?
Can the view buffer even be updated, I have seen no flag specifying it as DYNAMIC?

### #2Quat  Members

Posted 20 March 2012 - 09:59 AM

The effects framework does a Map or UpdateResource behind the scenes. It's been a while since I looked at the effect code, but it should flag a variable as "dirty" and only update the constant buffer if one of the values changed. So it is not really an abuse.

However, one example of an improvement is your "per-frame" constant buffer. You only need one of these and all your effects can share it. I use a header and always put the per-frame cb in slot 0. Then I use ID3DX11Effect::GetConstantBufferByName and set the per-frame CB for each effect; this is just a bind operation (when Apply() is called), as the per-frame CB is only updated once per-frame.

If you don't do this, then each effect will maintain its own per-frame cbuffer. This is still not likely to make a performance impact, but still.
-----Quat

### #3Asesh  Members

Posted 20 March 2012 - 06:28 PM

This is how it's done:

// Retrieve the handles to the constant buffers

//

m_pD3DXEffectEveryFrameConstBuffer = m_pD3DXEffect->GetConstantBufferByName("CBEveryFrame"); // Constant buffer that changes every frame

if(!m_pD3DXEffectEveryFrameConstBuffer->IsValid())

return E_FAIL;

update constant buffer:

D3D11_MAPPED_SUBRESOURCE oD3DMappedSubresource;

CHECK_COM(pD3DDeviceContext->Map(m_pD3DMaterialConstbuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &oD3DMappedSubresource)); // Get a pointer to the data in the constant buffer

::memcpy(oD3DMappedSubresource.pData, pMaterialConstantBuffer, sizeof(SMaterialConstantBuffer)); // Copy the constant buffers

pD3DDeviceContext->Unmap(m_pD3DMaterialConstbuffer, 0);

CHECK_COM(m_pD3DXEffectMaterialConstBuffer->SetConstantBuffer(m_pD3DMaterialConstbuffer)); // Set material constant buffer in the shader

Assuming the constant buffer was created dynamic usage flag. If it's created with default usage flag then you have to use UpdateSubresource method to do so

### #4osiris11  Members

Posted 21 March 2012 - 03:47 AM

@ Quat: Your answer sounds good: it's easy, would not require me to do too much changes, and would not need me to do obscure things.

@ Asesh: It's quite the opposite to what Quat suggested: You tell me to actually take everything in my own responsibility, and actually mapping the buffer. In this case, would I have to create the buffers myself? As I see it, the effect creates the declared buffers within it on compilation(?) and I can just access it with the GetConstantBufferFunction. Where can I make it's usage DYNAMIC? Do I have to create the buffer myself and replace it with the one the shader generated?

Actually, now I'm as wise as before: It seems to me I have two concurring opinions on this topic, and (of course I am tempted to use the easier one for the time being) still can't decide which option to settle with.
Are there any more opinions?

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.