T* BodySystemGPU<T>::getArray(BodyArray array)
{
assert(m_bInitialized);
T *hdata = 0;
T *ddata = 0;
cudaGraphicsResource *pgres = nullptr;
int currentReadHost = m_bUseSysMem ? m_currentRead : 0;
switch (array)
{
default:
case BODYSYSTEM_POSITION:
hdata = m_hPos[currentReadHost];
ddata = m_deviceData[0].dPos[m_currentRead];
if (m_bUsePBO)
{
pgres = m_pGRes[m_currentRead];
}
break;
case BODYSYSTEM_VELOCITY:
hdata = m_hVel;
ddata = m_deviceData[0].dVel;
break;
}
if (!m_bUseSysMem)
{
if (pgres)
{
checkCudaErrors(cudaGraphicsResourceSetMapFlags(pgres, cudaGraphicsMapFlagsReadOnly));
checkCudaErrors(cudaGraphicsMapResources(1, &pgres, 0));
size_t bytes;
checkCudaErrors(cudaGraphicsResourceGetMappedPointer((void **)&ddata, &bytes, pgres));
}
checkCudaErrors(cudaMemcpy(hdata, ddata,
m_numBodies*4*sizeof(T), cudaMemcpyDeviceToHost));
if (pgres)
{
checkCudaErrors(cudaGraphicsUnmapResources(1, &pgres, 0));
}
}
return hdata;
}
HRESULT DX11CustomRenderTarget::OnResize( ID3D11Device* pd3dDevice, UINT width, UINT height)
{
HRESULT hr = S_OK;
if (width == m_Width && height == m_Height) return hr;
OnD3D11DestroyDevice();
m_Width = width;
m_Height = height;
m_Targets = new ID3D11Texture2D*[m_uNumTargets];
m_TargetsRTV = new ID3D11RenderTargetView*[m_uNumTargets];
m_TargetsSRV = new ID3D11ShaderResourceView*[m_uNumTargets];
//creating targets
{
for (UINT i = 0; i < m_uNumTargets; i++)
{
D3D11_TEXTURE2D_DESC textureDesc;
ZeroMemory(&textureDesc, sizeof(D3D11_TEXTURE2D_DESC));
textureDesc.Width = m_Width;
textureDesc.Height = height;
textureDesc.MipLevels = 1;
textureDesc.ArraySize = 1;
textureDesc.Format = m_TextureFormats[i];
//TODO fix this for MSAA/CSAA
textureDesc.SampleDesc.Count = 1;
textureDesc.SampleDesc.Quality = 0;
textureDesc.Usage = D3D11_USAGE_DEFAULT;
textureDesc.BindFlags = D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE;
textureDesc.CPUAccessFlags = 0;
textureDesc.MiscFlags = 0;
D3D11_RENDER_TARGET_VIEW_DESC rtvDesc;
ZeroMemory(&rtvDesc, sizeof(D3D11_RENDER_TARGET_VIEW_DESC));
rtvDesc.Format = textureDesc.Format;
rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D;
rtvDesc.Texture2D.MipSlice = 0;
D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc;
ZeroMemory(&srvDesc, sizeof(D3D11_SHADER_RESOURCE_VIEW_DESC));
srvDesc.Format = textureDesc.Format;
srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
srvDesc.Texture2D.MipLevels = 1;
srvDesc.Texture2D.MostDetailedMip = 0;
V_RETURN(pd3dDevice->CreateTexture2D(&textureDesc, NULL, &m_Targets[i]));
V_RETURN(pd3dDevice->CreateRenderTargetView(m_Targets[i], &rtvDesc, &m_TargetsRTV[i]));
V_RETURN(pd3dDevice->CreateShaderResourceView(m_Targets[i], &srvDesc, &m_TargetsSRV[i]));
}
}
//creating depth stencil
{
D3D11_TEXTURE2D_DESC desc;
ZeroMemory(&desc, sizeof(D3D11_TEXTURE2D_DESC));
desc.Usage = D3D11_USAGE_DEFAULT;
desc.Format = DXGI_FORMAT_R32_TYPELESS;
desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_DEPTH_STENCIL;
desc.MipLevels = 1;
desc.ArraySize = 1;
desc.SampleDesc.Count = 1;
desc.SampleDesc.Quality = 0;
desc.CPUAccessFlags = 0;
desc.Height = height;
desc.Width = width;
V_RETURN(pd3dDevice->CreateTexture2D(&desc, NULL, &m_DepthStencil));
D3D11_DEPTH_STENCIL_VIEW_DESC descDSV;
ZeroMemory(&descDSV, sizeof(D3D11_DEPTH_STENCIL_VIEW_DESC));
descDSV.Format = DXGI_FORMAT_D32_FLOAT;
if (desc.SampleDesc.Count > 1) {
descDSV.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DMS;
} else {
descDSV.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D;
descDSV.Texture2D.MipSlice = 0;
}
V_RETURN(pd3dDevice->CreateDepthStencilView(m_DepthStencil, &descDSV, &m_DepthStencilDSV));
D3D11_SHADER_RESOURCE_VIEW_DESC descSRV;
ZeroMemory(&descSRV, sizeof(D3D11_SHADER_RESOURCE_VIEW_DESC));
descSRV.Format = DXGI_FORMAT_R32_FLOAT;
descSRV.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D;
descSRV.Texture2D.MipLevels = 1;
descSRV.Texture2D.MostDetailedMip = 0;
V_RETURN(pd3dDevice->CreateShaderResourceView(m_DepthStencil, &descSRV, &m_DepthStencilSRV));
}
if (m_dCudaFloat) {
for (UINT i = 0; i < m_uNumTargets; i++) {
cutilSafeCall(cudaGraphicsUnregisterResource(m_dCudaFloat[i]));
}
SAFE_DELETE_ARRAY(m_dCudaFloat)
}
m_dCudaFloat = new cudaGraphicsResource*[m_uNumTargets];
for (UINT i = 0; i < m_uNumTargets; i++) {
cutilSafeCall(cudaGraphicsD3D11RegisterResource(&m_dCudaFloat[i], m_Targets[i], cudaGraphicsRegisterFlagsNone));
cutilSafeCall(cudaGraphicsResourceSetMapFlags(m_dCudaFloat[i], cudaGraphicsMapFlagsReadOnly));
}
return hr;
}
cudaError_t WINAPI wine_cudaGraphicsResourceSetMapFlags( struct cudaGraphicsResource *resource, unsigned int flags ) {
WINE_TRACE("\n");
return cudaGraphicsResourceSetMapFlags( resource, flags );
}
inline void setMapFlags(unsigned int f)
{
flags = f;
CUDA_CHECK(cudaGraphicsResourceSetMapFlags(resource, flags));
}
