//[-------------------------------------------------------]
//[ Public methods ]
//[-------------------------------------------------------]
TextureBufferDsa::TextureBufferDsa(OpenGLRenderer &openGLRenderer, uint32_t numberOfBytes, Renderer::TextureFormat::Enum textureFormat, const void *data, Renderer::BufferUsage bufferUsage) :
TextureBuffer(openGLRenderer)
{
if (openGLRenderer.getExtensions().isGL_ARB_direct_state_access())
{
{ // Buffer part
// Create the OpenGL texture buffer
glCreateBuffers(1, &mOpenGLTextureBuffer);
// Upload the data
// -> Usage: These constants directly map to "GL_ARB_vertex_buffer_object" and OpenGL ES 2 constants, do not change them
glNamedBufferData(mOpenGLTextureBuffer, static_cast<GLsizeiptr>(numberOfBytes), data, static_cast<GLenum>(bufferUsage));
}
{ // Texture part
// Create the OpenGL texture instance
glCreateTextures(GL_TEXTURE_BUFFER_ARB, 1, &mOpenGLTexture);
// Attach the storage for the buffer object to the buffer texture
glTextureBuffer(mOpenGLTexture, Mapping::getOpenGLInternalFormat(textureFormat), mOpenGLTextureBuffer);
}
}
else
{
// Create the OpenGL texture buffer
glGenBuffersARB(1, &mOpenGLTextureBuffer);
// Create the OpenGL texture instance
glGenTextures(1, &mOpenGLTexture);
// Buffer part
// -> Upload the data
// -> Usage: These constants directly map to "GL_ARB_vertex_buffer_object" and OpenGL ES 2 constants, do not change them
glNamedBufferDataEXT(mOpenGLTextureBuffer, static_cast<GLsizeiptr>(numberOfBytes), data, static_cast<GLenum>(bufferUsage));
{ // Texture part
#ifndef OPENGLRENDERER_NO_STATE_CLEANUP
// Backup the currently bound OpenGL texture
GLint openGLTextureBackup = 0;
glGetIntegerv(GL_TEXTURE_BINDING_BUFFER_ARB, &openGLTextureBackup);
#endif
// Make this OpenGL texture instance to the currently used one
glBindTexture(GL_TEXTURE_BUFFER_ARB, mOpenGLTexture);
// Attaches the storage for the buffer object to the active buffer texture
// -> Sadly, there's no direct state access (DSA) function defined for this in "GL_EXT_direct_state_access"
glTexBufferARB(GL_TEXTURE_BUFFER_ARB, Mapping::getOpenGLInternalFormat(textureFormat), mOpenGLTextureBuffer);
#ifndef OPENGLRENDERER_NO_STATE_CLEANUP
// Be polite and restore the previous bound OpenGL texture
glBindTexture(GL_TEXTURE_BUFFER_ARB, static_cast<GLuint>(openGLTextureBackup));
#endif
}
}
}
//------------------------------------------------------------------------------
//
//------------------------------------------------------------------------------
bool RendererStandard::initResourcesGrid()
{
//
// Grid floor
//
glGenBuffers(1, &s_vboGrid);
vec3f *data = new vec3f[GRIDDEF * 4];
vec3f *p = data;
int j = 0;
for (int i = 0; i<GRIDDEF; i++)
{
*(p++) = vec3f(-GRIDSZ, 0.0, GRIDSZ*(-1.0f + 2.0f*(float)i / (float)GRIDDEF));
*(p++) = vec3f(GRIDSZ*(1.0f - 2.0f / (float)GRIDDEF), 0.0, GRIDSZ*(-1.0f + 2.0f*(float)i / (float)GRIDDEF));
*(p++) = vec3f(GRIDSZ*(-1.0f + 2.0f*(float)i / (float)GRIDDEF), 0.0, -GRIDSZ);
*(p++) = vec3f(GRIDSZ*(-1.0f + 2.0f*(float)i / (float)GRIDDEF), 0.0, GRIDSZ*(1.0f - 2.0f / (float)GRIDDEF));
}
s_vboGridSz = sizeof(vec3f)*GRIDDEF * 4;
glNamedBufferDataEXT(s_vboGrid, s_vboGridSz, data[0].vec_array, GL_STATIC_DRAW);
delete[] data;
//
// Target Cross
//
glGenBuffers(1, &s_vboCross);
vec3f crossVtx[6] = {
vec3f(-CROSSSZ, 0.0f, 0.0f), vec3f(CROSSSZ, 0.0f, 0.0f),
vec3f(0.0f, -CROSSSZ, 0.0f), vec3f(0.0f, CROSSSZ, 0.0f),
vec3f(0.0f, 0.0f, -CROSSSZ), vec3f(0.0f, 0.0f, CROSSSZ),
};
s_vboCrossSz = sizeof(vec3f)* 6;
glNamedBufferDataEXT(s_vboCross, s_vboCrossSz, crossVtx[0].vec_array, GL_STATIC_DRAW);
return true;
}
bool initArrayBuffer()
{
glGenBuffers(buffer::MAX, BufferName);
glNamedBufferDataEXT(BufferName[buffer::ELEMENT], ElementSize, ElementData, GL_STATIC_DRAW);
glNamedBufferDataEXT(BufferName[buffer::VERTEX], VertexSize, VertexData, GL_STATIC_DRAW);
return glf::checkError("initArrayBuffer");;
}
bool initBuffer()
{
glGenBuffers(buffer::MAX, BufferName);
glNamedBufferDataEXT(BufferName[buffer::VERTEX], VertexSize, VertexData, GL_STATIC_DRAW);
glNamedBufferDataEXT(BufferName[buffer::TRANSFORM], sizeof(glm::mat4), NULL, GL_DYNAMIC_DRAW);
glNamedBufferDataEXT(BufferName[buffer::BLIT], sizeof(glm::mat4), NULL, GL_DYNAMIC_DRAW);
return true;
}
bool initBuffer()
{
GLint UniformBufferOffset(0);
glGetIntegerv(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &UniformBufferOffset);
UniformBlockSize = glm::max(GLint(sizeof(glm::mat4)), UniformBufferOffset);
glGenBuffers(buffer::MAX, &BufferName[0]);
glNamedBufferDataEXT(BufferName[buffer::ELEMENT], ElementSize, ElementData, GL_STATIC_DRAW);
glNamedBufferDataEXT(BufferName[buffer::VERTEX], VertexSize, VertexData, GL_STATIC_DRAW);
glNamedBufferDataEXT(BufferName[buffer::TRANSFORM], UniformBlockSize, nullptr, GL_DYNAMIC_DRAW);
return true;
}
OGLGraphicsBuffer::OGLGraphicsBuffer(BufferUsage usage, uint32_t access_hint, GLenum target, ElementInitData const * init_data)
: GraphicsBuffer(usage, access_hint),
target_(target)
{
BOOST_ASSERT((GL_ARRAY_BUFFER == target) || (GL_ELEMENT_ARRAY_BUFFER == target));
glGenBuffers(1, &vb_);
if (init_data != nullptr)
{
size_in_byte_ = init_data->row_pitch;
if (glloader_GL_EXT_direct_state_access())
{
glNamedBufferDataEXT(vb_, static_cast<GLsizeiptr>(size_in_byte_), init_data->data,
(BU_Static == usage_) ? GL_STATIC_DRAW : GL_DYNAMIC_DRAW);
}
else
{
OGLRenderEngine& re = *checked_cast<OGLRenderEngine*>(&Context::Instance().RenderFactoryInstance().RenderEngineInstance());
re.BindBuffer(target_, vb_);
glBufferData(target_,
static_cast<GLsizeiptr>(size_in_byte_), init_data->data,
(BU_Static == usage_) ? GL_STATIC_DRAW : GL_DYNAMIC_DRAW);
}
}
}
void RMesh_CreateTexCoords (model_t *mod, dmdl_t *hdr)
{
mdlst_t *st = (mdlst_t *) Scratch_Alloc ();
int *order = (int *) ((byte *) hdr + hdr->ofs_glcmds);
mod->numframeverts = 0;
glGenBuffers (1, &mod->texcoordvbo);
while (1)
{
// get the vertex count and primitive type
int count = *order++;
int i;
if (!count) break;
if (count < 0) count = -count;
for (i = 0; i < count; i++, st++, order += 3)
{
st->texcoords[0] = ((float *) order)[0];
st->texcoords[1] = ((float *) order)[1];
}
mod->numframeverts += count;
}
glNamedBufferDataEXT (mod->texcoordvbo, mod->numframeverts * sizeof (mdlst_t), Scratch_Alloc (), GL_STATIC_DRAW);
}
void RMesh_CreateFrames (model_t *mod, dmdl_t *hdr)
{
int i, frame;
posevert_t *verts = (posevert_t *) Scratch_Alloc ();
int numverts = 0;
glGenBuffers (1, &mod->meshvbo);
for (frame = 0; frame < hdr->num_frames; frame++)
{
int *order = (int *) ((byte *) hdr + hdr->ofs_glcmds);
dtrivertx_t *vert = ((daliasframe_t *) ((byte *) hdr + hdr->ofs_frames + frame * hdr->framesize))->verts;
while (1)
{
// get the vertex count and primitive type
int count = *order++;
if (!count) break;
if (count < 0) count = -count;
for (i = 0; i < count; i++, verts++, order += 3)
{
verts->position[0] = vert[order[2]].v[0];
verts->position[1] = vert[order[2]].v[1];
verts->position[2] = vert[order[2]].v[2];
verts->position[3] = vert[order[2]].lightnormalindex;
}
numverts += count;
}
}
glNamedBufferDataEXT (mod->meshvbo, numverts * sizeof (posevert_t), Scratch_Alloc (), GL_STATIC_DRAW);
}
void RMesh_CreatePrograms (void)
{
glGetIntegerv (GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT, &gl_meshuboblocksize);
if (gl_meshuboblocksize < sizeof (meshubo_t))
{
int addsize = gl_meshuboblocksize;
while (gl_meshuboblocksize < sizeof (meshubo_t))
gl_meshuboblocksize += addsize;
}
gl_meshprog = GL_CreateShaderFromName ("glsl/mesh.glsl", "MeshVS", "MeshFS");
glUniformBlockBinding (gl_meshprog, glGetUniformBlockIndex (gl_meshprog, "MeshUniforms"), gl_meshubobinding);
glProgramUniform1i (gl_meshprog, glGetUniformLocation (gl_meshprog, "diffuse"), 0);
glProgramUniform3fv (gl_meshprog, glGetUniformLocation (gl_meshprog, "lightnormal"), 162, (float *) r_avertexnormals);
u_meshMaxLights = glGetUniformLocation(gl_meshprog, "r_maxLights");
u_meshEntOrig = glGetUniformLocation(gl_meshprog, "r_entOrig");
for (int i = 0; i < MAX_LIGHTS; ++i)
{
u_meshLightPos[i] = glGetUniformLocation(gl_meshprog, va("Lights.origin[%i]", i));
u_meshLightColor[i] = glGetUniformLocation(gl_meshprog, va("Lights.color[%i]", i));
u_meshLightAtten[i] = glGetUniformLocation(gl_meshprog, va("Lights.radius[%i]", i));
}
glGenBuffers (1, &gl_meshubo);
glNamedBufferDataEXT (gl_meshubo, MESH_UBO_MAX_BLOCKS * gl_meshuboblocksize, NULL, GL_STREAM_DRAW);
}
void SObjModel::BindVAOs() {
/* load descriptor*/
//unsigned int temp_vao;
unsigned int temp_vbo;
unsigned int temp_ibo;
/*
std::string map_Ka; ambient map
std::string map_Kd; diffuse map
std::string map_Ns; Specular shinnes map
std::string map_d; alpha map
std::string map_bump;
*/
/*https://devtalk.nvidia.com/default/topic/561172/gldrawarrays-without-vao-for-procedural-geometry-using-gl_vertexid-doesn-t-work-in-debug-context/ */
glGenVertexArrays ( 1, &d_emptyVAO );
// glBindVertexArray(d_emptyVAO);
for (auto it = d_sm.begin(); it != d_sm.end();++it) {
auto &submesh = (*it);
/*gen buffers for submesh*/
/*vertixes*/
temp_vbo = 0;
temp_ibo = 0;
glGenBuffers ( 1, &temp_vbo );
glGenBuffers(1, &temp_ibo);
//MASSERT(submesh->vn.empty());
// MASSERT(submesh->indexes.empty());
//gl 4.5 without EXT
#ifndef __APPLE__
glNamedBufferDataEXT(temp_vbo,submesh->vn.size() *sizeof(UVNVertex), submesh->vn.data(), GL_STATIC_DRAW);
glNamedBufferDataEXT(temp_ibo,submesh->indexes.size() *sizeof(unsigned int), submesh->indexes.data(), GL_STATIC_DRAW);
SubMeshIDs idx;
idx.vbo = temp_vbo;
idx.ibo = temp_ibo;
submesh_idx[submesh->id] = idx;
#else
//todo
#endif
glBindBuffer(GL_ARRAY_BUFFER,0);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER,0);
}
//glBindVertexArray(0);
}
void Buffer::setData(GLsizeiptr size, const GLvoid* data, GLenum usage)
{
if (m_directStateAccess)
{
glNamedBufferDataEXT(m_id, size, data, usage);
}
else
{
bind();
glBufferData(m_target, size, data, usage);
CheckGLError();
}
}
bool
CLGLVertexBuffer::allocate(cl_context clContext) {
assert(clContext);
// create GL buffer first
int size = _numElements * _numVertices * sizeof(float);
glGenBuffers(1, &_vbo);
#if defined(GL_EXT_direct_state_access)
if (glNamedBufferDataEXT) {
glNamedBufferDataEXT(_vbo, size, 0, GL_DYNAMIC_DRAW);
} else {
#else
{
#endif
GLint prev = 0;
glGetIntegerv(GL_ARRAY_BUFFER_BINDING, &prev);
glBindBuffer(GL_ARRAY_BUFFER, _vbo);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, prev);
}
// register vbo as cl memory
cl_int err;
_clMemory = clCreateFromGLBuffer(clContext,
CL_MEM_READ_WRITE, _vbo, &err);
if (err != CL_SUCCESS) return false;
return true;
}
void
CLGLVertexBuffer::map(cl_command_queue queue) {
if (_clMapped) return; // XXX: what if another queue is given?
_clQueue = queue;
clEnqueueAcquireGLObjects(queue, 1, &_clMemory, 0, 0, 0);
_clMapped = true;
}
void
CLGLVertexBuffer::unmap() {
if (! _clMapped) return;
clEnqueueReleaseGLObjects(_clQueue, 1, &_clMemory, 0, 0, 0);
_clMapped = false;
}
} // end namespace Osd
void OGLGraphicsBuffer::DoResize()
{
BOOST_ASSERT(size_in_byte_ != 0);
if (glloader_GL_EXT_direct_state_access())
{
glNamedBufferDataEXT(vb_, static_cast<GLsizeiptr>(size_in_byte_), nullptr,
(BU_Static == usage_) ? GL_STATIC_DRAW : GL_DYNAMIC_DRAW);
}
else
{
OGLRenderEngine& re = *checked_cast<OGLRenderEngine*>(&Context::Instance().RenderFactoryInstance().RenderEngineInstance());
re.BindBuffer(target_, vb_);
glBufferData(target_,
static_cast<GLsizeiptr>(size_in_byte_), nullptr,
(BU_Static == usage_) ? GL_STATIC_DRAW : GL_DYNAMIC_DRAW);
}
}
void RMesh_CreateIndexes (model_t *mod, dmdl_t *hdr)
{
unsigned short *ndx = (unsigned short *) Scratch_Alloc ();
int *order = (int *) ((byte *) hdr + hdr->ofs_glcmds);
int firstvertex = 0;
mod->numindexes = 0;
glGenBuffers (1, &mod->indexbuffer);
while (1)
{
// get the vertex count and primitive type
int count = *order++;
int i;
if (!count) break;
if (count < 0)
{
for (i = 2, count = -count; i < count; i++, ndx += 3)
{
ndx[0] = firstvertex + 0;
ndx[1] = firstvertex + (i - 1);
ndx[2] = firstvertex + i;
}
}
else
{
for (i = 2; i < count; i++, ndx += 3)
{
ndx[0] = firstvertex + i - 2;
ndx[1] = firstvertex + ((i & 1) ? i : (i - 1));
ndx[2] = firstvertex + ((i & 1) ? (i - 1) : i);
}
}
order += count * 3;
firstvertex += count;
mod->numindexes += (count - 2) * 3;
}
glNamedBufferDataEXT (mod->indexbuffer, mod->numindexes * sizeof (unsigned short), Scratch_Alloc (), GL_STATIC_DRAW);
}
bool
CudaGLVertexBuffer::allocate() {
int size = _numElements * _numVertices * sizeof(float);
glGenBuffers(1, &_vbo);
#if defined(GL_EXT_direct_state_access)
if (glNamedBufferDataEXT) {
glNamedBufferDataEXT(_vbo, size, 0, GL_DYNAMIC_DRAW);
} else {
#else
{
#endif
glBindBuffer(GL_ARRAY_BUFFER, _vbo);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, 0);
}
// register vbo as cuda resource
cudaError_t err = cudaGraphicsGLRegisterBuffer(
&_cudaResource, _vbo, cudaGraphicsMapFlagsWriteDiscard);
if (err != cudaSuccess) return false;
return true;
}
void
CudaGLVertexBuffer::map() {
if (_devicePtr) return;
size_t num_bytes;
void *ptr;
cudaError_t err = cudaGraphicsMapResources(1, &_cudaResource, 0);
if (err != cudaSuccess)
Far::Error(Far::FAR_RUNTIME_ERROR,
"CudaGLVertexBuffer::map failed.\n%s\n",
cudaGetErrorString(err));
err = cudaGraphicsResourceGetMappedPointer(&ptr, &num_bytes, _cudaResource);
if (err != cudaSuccess)
Far::Error(Far::FAR_RUNTIME_ERROR,
"CudaGLVertexBuffer::map failed.\n%s\n",
cudaGetErrorString(err));
_devicePtr = ptr;
}
void
CudaGLVertexBuffer::unmap() {
if (_devicePtr == NULL) return;
cudaError_t err = cudaGraphicsUnmapResources(1, &_cudaResource, 0);
if (err != cudaSuccess)
Far::Error(Far::FAR_RUNTIME_ERROR,
"CudaGLVertexBuffer::unmap failed.\n%s\n",
cudaGetErrorString(err));
_devicePtr = NULL;
}
} // end namespace Osd
void Buffer::BufferData(GLsizeiptr size, const GLvoid* data, GLenum usage)
{
Generate();
glNamedBufferDataEXT(m_handle, size, data, usage);
m_size = size;
}
//------------------------------------------------------------------------------
// It is possible to create one VBO for one Mesh; and one EBO for each primitive group
// however, for this sample, we will create only one VBO for all and one EBO
// meshes and primitive groups will have an offset in these buffers
//------------------------------------------------------------------------------
bool Bk3dModelStandard::initResourcesObject()
{
LOGFLUSH();
SHOWPROGRESS("Init resources")
//m_pGenericModel->m_meshFile->pMeshes->n = 60000;
//
// create Buffer Object for materials
//
if(m_pGenericModel->m_meshFile->pMaterials && m_pGenericModel->m_meshFile->pMaterials->nMaterials )
{
//
// Material UBO: *TABLE* of multiple materials
// Then offset in it for various drawcalls
//
if(m_uboMaterial.Id == 0)
glGenBuffers(1, &m_uboMaterial.Id);
m_uboMaterial.Sz = sizeof(MaterialBuffer) * m_pGenericModel->m_materialNItems;
glNamedBufferDataEXT(m_uboMaterial.Id, m_uboMaterial.Sz, m_pGenericModel->m_material, GL_STATIC_DRAW);
glBindBufferBase(GL_UNIFORM_BUFFER,UBO_MATERIAL, m_uboMaterial.Id);
LOGI("%d materials stored in %d Kb\n", m_pGenericModel->m_meshFile->pMaterials->nMaterials, (m_uboMaterial.Sz+512)/1024);
LOGFLUSH();
}
//
// create Buffer Object for Object-matrices
//
if(m_pGenericModel->m_meshFile->pTransforms && m_pGenericModel->m_meshFile->pTransforms->nBones)
{
//
// Transformation UBO: *TABLE* of multiple transformations
// Then offset in it for various drawcalls
//
if(m_uboObjectMatrices.Id == 0)
glGenBuffers(1, &m_uboObjectMatrices.Id);
m_uboObjectMatrices.Sz = sizeof(MatrixBufferObject) * m_pGenericModel->m_objectMatricesNItems;
glNamedBufferDataEXT(m_uboObjectMatrices.Id, m_uboObjectMatrices.Sz, m_pGenericModel->m_objectMatrices, GL_STATIC_DRAW);
glBindBufferBase(GL_UNIFORM_BUFFER,UBO_MATRIXOBJ, m_uboObjectMatrices.Id);
LOGI("%d matrices stored in %d Kb\n", m_pGenericModel->m_meshFile->pTransforms->nBones, (m_uboObjectMatrices.Sz + 512)/1024);
LOGFLUSH();
}
//
// First pass: evaluate the size of the single VBO
// and store offset to where we'll find data back
//
bk3d::Mesh *pMesh = NULL;
for(int i=0; i< m_pGenericModel->m_meshFile->pMeshes->n; i++)
{
SETPROGRESSVAL(100.0f*(float)i/(float)m_pGenericModel->m_meshFile->pMeshes->n);
pMesh = m_pGenericModel->m_meshFile->pMeshes->p[i];
//
// Slots: buffers for vertices
//
int n = pMesh->pSlots->n;
for(int s=0; s<n; s++)
{
bk3d::Slot* pS = pMesh->pSlots->p[s];
GLuint id;
glGenBuffers(1, &id); // Buffer Object directly kept in the Slot
pS->userData = id;
#if 1
glNamedBufferDataEXT(id, pS->vtxBufferSizeBytes, NULL, GL_STATIC_DRAW);
#else
glNamedBufferStorageEXT(id, pS->vtxBufferSizeBytes, NULL, 0); // Not working with NSight !!! https://www.opengl.org/registry/specs/ARB/buffer_storage.txt
#endif
glNamedBufferSubDataEXT(id, 0, pS->vtxBufferSizeBytes, pS->pVtxBufferData);
}
//
// Primitive groups
//
for(int pg=0; pg<pMesh->pPrimGroups->n; pg++)
{
bk3d::PrimGroup* pPG = pMesh->pPrimGroups->p[pg];
if(pPG->indexArrayByteSize > 0)
{
if((pPG->pOwnerOfIB == pPG)||(pPG->pOwnerOfIB == NULL)) // this primitive group doesn't use other's buffer
{
GLuint id;
glGenBuffers(1, &id);
pPG->userPtr = (int*)id;
#if 1
glNamedBufferDataEXT(id, pPG->indexArrayByteSize, NULL, GL_STATIC_DRAW);
#else
glNamedBufferStorageEXT(id, pPG->indexArrayByteSize, NULL, 0); // Not working with NSight !!! https://www.opengl.org/registry/specs/ARB/buffer_storage.txt
#endif
glNamedBufferSubDataEXT(id, pPG->indexArrayByteOffset, pPG->indexArrayByteSize, pPG->pIndexBufferData);
} else {
pPG->userPtr = pPG->pOwnerOfIB->userPtr;
}
} else {
pPG->userPtr = NULL;
}
}
}
//LOGI("meshes: %d in :%d VBOs (%f Mb) and %d EBOs (%f Mb) \n", m_pGenericModel->m_meshFile->pMeshes->n, .size(), (float)totalVBOSz/(float)(1024*1024), m_ObjEBOs.size(), (float)totalEBOSz/(float)(1024*1024));
LOGFLUSH();
HIDEPROGRESS()
return true;
}
//------------------------------------------------------------------------------
//
//------------------------------------------------------------------------------
bool RendererStandard::initGraphics(int w, int h, int MSAA)
{
//
// some offscreen buffer
//
fboInitialize(w, h, MSAA);
//
// Shader compilation
//
if(!s_shaderGrid.addVertexShaderFromString(g_glslv_grid))
return false;
if(!s_shaderGrid.addFragmentShaderFromString(g_glslf_grid))
return false;
if (!s_shaderGrid.link())
return false;
if(!s_shaderMesh.addVertexShaderFromString(s_glslv_mesh))
return false;
if(!s_shaderMesh.addFragmentShaderFromString(s_glslf_mesh))
return false;
if(!s_shaderMesh.link())
return false;
if(!s_shaderMeshLine.addVertexShaderFromString(s_glslv_mesh))
return false;
if(!s_shaderMeshLine.addFragmentShaderFromString(s_glslf_mesh_line))
return false;
if(!s_shaderMeshLine.link())
return false;
//
// Create some UBO for later share their 64 bits
//
glGenBuffers(1, &g_uboMatrix.Id);
g_uboMatrix.Sz = sizeof(MatrixBufferGlobal);
glNamedBufferDataEXT(g_uboMatrix.Id, g_uboMatrix.Sz, &g_globalMatrices, GL_STREAM_DRAW);
//glBindBufferBase(GL_UNIFORM_BUFFER,UBO_MATRIX, g_uboMatrix.Id);
//
// Trivial Light info...
//
glGenBuffers(1, &g_uboLight.Id);
g_uboLight.Sz = sizeof(LightBuffer);
glNamedBufferDataEXT(g_uboLight.Id, g_uboLight.Sz, &s_light, GL_STATIC_DRAW);
//glBindBufferBase(GL_UNIFORM_BUFFER,UBO_LIGHT, g_uboLight.Id);
//
// Misc OGL setup
//
glClearColor(0.0f, 0.1f, 0.15f, 1.0f);
glGenVertexArrays(1, &s_vao);
glBindVertexArray(s_vao);
//
// Grid
//
if(!initResourcesGrid())
return false;
//
// OpenGL timers
//
m_gltimers.init(10);
LOGOK("Initialized renderer %s\n", getName());
m_bValid = true;
return true;
}
void
Hd_SmoothNormalsComputationGPU::Execute(
HdBufferArrayRangeSharedPtr const &range)
{
HD_TRACE_FUNCTION();
HF_MALLOC_TAG_FUNCTION();
if (!glDispatchCompute)
return;
TF_VERIFY(_adjacency);
HdBufferArrayRangeSharedPtr const &adjacencyRange = _adjacency->GetAdjacencyRange();
TF_VERIFY(adjacencyRange);
// select shader by datatype
TfToken shaderToken;
if (_srcDataType == GL_FLOAT) {
if (_dstDataType == GL_FLOAT) {
shaderToken = HdGLSLProgramTokens->smoothNormalsFloatToFloat;
} else if (_dstDataType == GL_DOUBLE) {
shaderToken = HdGLSLProgramTokens->smoothNormalsFloatToDouble;
} else if (_dstDataType == GL_INT_2_10_10_10_REV) {
shaderToken = HdGLSLProgramTokens->smoothNormalsFloatToPacked;
}
} else if (_srcDataType == GL_DOUBLE) {
if (_dstDataType == GL_FLOAT) {
shaderToken = HdGLSLProgramTokens->smoothNormalsDoubleToFloat;
} else if (_dstDataType == GL_DOUBLE) {
shaderToken = HdGLSLProgramTokens->smoothNormalsDoubleToDouble;
} else if (_dstDataType == GL_INT_2_10_10_10_REV) {
shaderToken = HdGLSLProgramTokens->smoothNormalsDoubleToPacked;
}
}
if (!TF_VERIFY(!shaderToken.IsEmpty())) return;
HdGLSLProgramSharedPtr computeProgram
= HdGLSLProgram::GetComputeProgram(shaderToken);
if (!computeProgram) return;
GLuint program = computeProgram->GetProgram().GetId();
// buffer resources for GPU computation
HdBufferResourceSharedPtr points = range->GetResource(_srcName);
HdBufferResourceSharedPtr normals = range->GetResource(_dstName);
HdBufferResourceSharedPtr adjacency = adjacencyRange->GetResource();
// prepare uniform buffer for GPU computation
struct Uniform {
int vertexOffset;
int adjacencyOffset;
int pointsOffset;
int pointsStride;
int normalsOffset;
int normalsStride;
} uniform;
// coherent vertex offset in aggregated buffer array
uniform.vertexOffset = range->GetOffset();
// adjacency offset/stride in aggregated adjacency table
uniform.adjacencyOffset = adjacencyRange->GetOffset();
// interleaved offset/stride to points
// note: this code (and the glsl smooth normal compute shader) assumes
// components in interleaved vertex array are always same data type.
// i.e. it can't handle an interleaved array which interleaves
// float/double, float/int etc.
uniform.pointsOffset = points->GetOffset() / points->GetComponentSize();
uniform.pointsStride = points->GetStride() / points->GetComponentSize();
// interleaved offset/stride to normals
uniform.normalsOffset = normals->GetOffset() / normals->GetComponentSize();
uniform.normalsStride = normals->GetStride() / normals->GetComponentSize();
// The number of points is based off the size of the output,
// However, the number of points in the adjacency table
// is computed based off the largest vertex indexed from
// to topology (aka topology->ComputeNumPoints).
//
// Therefore, we need to clamp the number of points
// to the number of entries in the adjancency table.
int numDestPoints = range->GetNumElements();
int numSrcPoints = _adjacency->GetNumPoints();
int numPoints = std::min(numSrcPoints, numDestPoints);
// transfer uniform buffer
GLuint ubo = computeProgram->GetGlobalUniformBuffer().GetId();
HdRenderContextCaps const &caps = HdRenderContextCaps::GetInstance();
// XXX: workaround for 319.xx driver bug of glNamedBufferDataEXT on UBO
// XXX: move this workaround to renderContextCaps
if (false && caps.directStateAccessEnabled) {
glNamedBufferDataEXT(ubo, sizeof(uniform), &uniform, GL_STATIC_DRAW);
} else {
glBindBuffer(GL_UNIFORM_BUFFER, ubo);
glBufferData(GL_UNIFORM_BUFFER, sizeof(uniform), &uniform, GL_STATIC_DRAW);
glBindBuffer(GL_UNIFORM_BUFFER, 0);
}
glBindBufferBase(GL_UNIFORM_BUFFER, 0, ubo);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, points->GetId());
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, normals->GetId());
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, adjacency->GetId());
// dispatch compute kernel
glUseProgram(program);
glDispatchCompute(numPoints, 1, 1);
glUseProgram(0);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
glBindBufferBase(GL_UNIFORM_BUFFER, 0, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, 0);
}
void BufferImplementation_DirectStateAccessEXT::setData(const Buffer * buffer, GLsizeiptr size, const GLvoid * data, GLenum usage) const
{
glNamedBufferDataEXT(buffer->id(), size, data, usage);
}
void
Hd_SmoothNormalsComputationGPU::Execute(
HdBufferArrayRangeSharedPtr const &range)
{
HD_TRACE_FUNCTION();
HD_MALLOC_TAG_FUNCTION();
if (not glDispatchCompute)
return;
// XXX: workaround until the shading stuff is implemeted.
// The drawing program is owned and set by testHdBasicDrawing now,
// so it has to be restored if it's changed in hd.
GLint restoreProgram = 0;
glGetIntegerv(GL_CURRENT_PROGRAM, &restoreProgram);
int numPoints = range->GetNumElements();
TF_VERIFY(_adjacency);
HdBufferArrayRangeSharedPtr const &adjacencyRange = _adjacency->GetAdjacencyRange();
TF_VERIFY(adjacencyRange);
// select shader by datatype
TfToken shaderToken = (_dstDataType == GL_FLOAT ?
HdGLSLProgramTokens->smoothNormalsFloat :
HdGLSLProgramTokens->smoothNormalsDouble);
HdGLSLProgramSharedPtr computeProgram = HdGLSLProgram::GetComputeProgram(shaderToken);
if (not computeProgram) return;
GLuint program = computeProgram->GetProgram().GetId();
// buffer resources for GPU computation
HdBufferResourceSharedPtr points = range->GetResource(_srcName);
HdBufferResourceSharedPtr normals = range->GetResource(_dstName);
HdBufferResourceSharedPtr adjacency = adjacencyRange->GetResource();
// prepare uniform buffer for GPU computation
struct Uniform {
int vertexOffset;
int adjacencyStride;
int adjacencyOffset;
int padding;
int pointsOffset;
int pointsStride;
int normalsOffset;
int normalsStride;
} uniform;
// coherent vertex offset in aggregated buffer array
uniform.vertexOffset = range->GetOffset();
// adjacency offset/stride in aggregated adjacency table
uniform.adjacencyStride = _adjacency->GetStride();
uniform.adjacencyOffset = adjacencyRange->GetOffset();
uniform.padding = 0;
// interleaved offset/stride to points
// note: this code (and the glsl smooth normal compute shader) assumes
// components in interleaved vertex array are always same data type.
// i.e. it can't handle an interleaved array which interleaves
// float/double, float/int etc.
uniform.pointsOffset = points->GetOffset() / points->GetComponentSize();
uniform.pointsStride = points->GetStride() / points->GetComponentSize();
// interleaved offset/stride to normals
uniform.normalsOffset = normals->GetOffset() / points->GetComponentSize();
uniform.normalsStride = normals->GetStride() / points->GetComponentSize();
// transfer uniform buffer
GLuint ubo = computeProgram->GetGlobalUniformBuffer().GetId();
HdRenderContextCaps const &caps = HdRenderContextCaps::GetInstance();
// XXX: workaround for 319.xx driver bug of glNamedBufferDataEXT on UBO
// XXX: move this workaround to renderContextCaps
if (false and caps.directStateAccessEnabled) {
glNamedBufferDataEXT(ubo, sizeof(uniform), &uniform, GL_STATIC_DRAW);
} else {
glBindBuffer(GL_UNIFORM_BUFFER, ubo);
glBufferData(GL_UNIFORM_BUFFER, sizeof(uniform), &uniform, GL_STATIC_DRAW);
glBindBuffer(GL_UNIFORM_BUFFER, 0);
}
glBindBufferBase(GL_UNIFORM_BUFFER, 0, ubo);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, points->GetId());
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, normals->GetId());
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, adjacency->GetId());
// dispatch compute kernel
glUseProgram(program);
glDispatchCompute(numPoints, 1, 1);
glUseProgram(0);
glBindBufferBase(GL_UNIFORM_BUFFER, 0, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, 0);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, 0);
// XXX: workaround until shading stuff implemeted.
glUseProgram(restoreProgram);
}
