Hey all, I have a few questions about mapping a dynamic vertex buffer in DirectX11

I currently have this struct that represents a Quad. This will hold the vertices of the quad, the texture, etc. I also have a Vector that will hold all of the Quads that should be drawn

struct Quad
{
    Vertex vertices[4];
};

std::vector<Quad> drawData;

How can I map all of the data in the drawData Vector to my Vertex buffer?

I have this but it will only draw one Quad (It keeps overwriting the data placed)

D3D11_MAPPED_SUBRESOURCE mapResource;
ZeroMemory(&mapResource, sizeof(mapResource));

batchContext->Map(vertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mapResource);

for(std::vector<Quad>::iterator i = drawData.begin(); i != drawData.end(); i++)
{
     memcpy(mapResource.pData, (*i).vertices, sizeof((*i).vertices));
}

batchContext->Unmap(vertexBuffer, 0);

How can I do something like

D3D11_MAPPED_SUBRESOURCE mapResource;
ZeroMemory(&mapResource, sizeof(mapResource));
int quadOffset = 0;

batchContext->Map(vertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mapResource);

for(std::vector<Quad>::iterator i = drawData.begin(); i != drawData.end(); i++)
{
     //Add the Quad onto mapResource/vertex buffer. Place it at the offset mark
     memcpy(mapResource.pData + quadOffset, (*i).vertices, sizeof((*i).vertices));
     
     //Add on 4 because 4 vertices to quad
     quadOffset += 4;
}

batchContext->Unmap(vertexBuffer, 0);

[s]
//Add on 4 because 4 vertices to quad
quadOffset += 4;
[/s]

This is just incrementing the pointer by 4 bytes each loop. Unless each Vertex is a single byte, that won't work. You could use...

[s]
quadoffset += sizeof(Quad);
[/s]

I guess that would work. But it might be simpler to do away with the loop entirely. Vectors are stored contiguously, so you can just memcpy it once, starting with the address of the first element, and with a copy size of the number of the vector elements * element size. So I think you could do...

D3D11_MAPPED_SUBRESOURCE mapResource;
ZeroMemory(&mapResource, sizeof(mapResource));

batchContext->Map(vertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mapResource);
 
//get byte length of all vector elements
size_t copySize = sizeof(Quad) * drawData.size();
 
//copy all elements at once
memcpy(mapResource.pData, &drawData[0], copySize);


batchContext->Unmap(vertexBuffer, 0);

EDIT: As pointed out below by Mona2000, if you need to keep the loop you have to cast the void pointer to Vertex* for your original offset+=4 to work...

for(std::vector<Quad>::iterator i = drawData.begin(); i != drawData.end(); i++)
{
     //Add the Quad onto mapResource/vertex buffer. Place it at the offset mark
     memcpy((Vertex*)mapResource.pData + quadOffset, (*i).vertices, sizeof((*i).vertices));
     
     //Add on 4 because 4 vertices to quad
     quadOffset += 4;
}

//Add on 4 because 4 vertices to quad
quadOffset += 4;

This is just incrementing the pointer by 4 bytes each loop. Unless each Vertex is a single byte, that won't work. You could use...

It's actually doing nothing because you can't do pointer arithmetics on a void * (that code won't even compile). It will work if pData is cast to Vertex * before adding the offset.

//Add on 4 because 4 vertices to quad
quadOffset += 4;

This is just incrementing the pointer by 4 bytes each loop. Unless each Vertex is a single byte, that won't work. You could use...

It's actually doing nothing because you can't do pointer arithmetics on a void * (that code won't even compile). It will work if pData is cast to Vertex * before adding the offset.

That's a very good point, you do need to cast the pData to a type to do any pointer arithmetic on it. My bad, thanks for clarifying.

It's actually doing nothing because you can't do pointer arithmetics on a void * (that code won't even compile). It will work if pData is cast to Vertex * before adding the offset.

quadOffset += 4;

You are absolutely right. I just didn't know how to explain it any other way

@Everyone

I guess I should explian a little more

My idea is to create a large vertex buffer, for example a buffer that can hold 10000 quads.

Where the flag changes between DISCARD and NOOVERWRITE depending on how much space I have left in my buffer. The only part

I'm not too sure on is actually getting the Quad data into the right spot so that when I call

deviceContext->Draw(vertCount,0);

I draw all of the quads instead of just one, because I keep overwriting the same data space on each pass of the for loop

So I was thinking something like

D3D11_MAPPED_SUBRESOURCE mapResource;
ZeroMemory(&mapResource, sizeof(mapResource));

int quadOffset = 0;
D3D11_Mapflag flag = D3D11_MAP_WRITE_NO_OVERWRITE;

batchContext->Map(vertexBuffer, 0, flag, 0, &mapResource);

for(std::vector<Quad>::iterator i = drawData.begin(); i != drawData.end(); i++)
{
    if(/*Texture needs to change or hit the buffer limit*/)
    {
        batchContext->Unmap(vertexBuffer, 0);
		
        /* Draw call and related code*/
		
        if(/*buffer was at the limit*/)
			flag = D3D11_MAP_WRITE_DISCARD;
        
		batchContext->Map(vertexBuffer, 0, flag, 0, &mapResource);
		flag = D3D11_MAP_WRITE_NO_OVERWRITE;
    }

	/*
	The code that places the vertices of the current quad onto into the pData mapResource.
	Then adds on an offset amount to let pData know we are going ot use the next 4 vertice spots
	*/
	
}

batchContext->Unmap(vertexBuffer, 0);

Its just that single part where I'm not too sure how to do it