// --------------------------------------------------------------------------------------------------------------------
bool UntexturedObjectsGLBindless::Init(const std::vector<UntexturedObjectsProblem::Vertex>& _vertices,
const std::vector<UntexturedObjectsProblem::Index>& _indices,
size_t _objectCount)
{
if (glBufferStorage == nullptr) {
console::warn("Unable to initialize solution '%s', glBufferStorage() unavailable.", GetName().c_str());
return false;
}
if (glGetBufferParameterui64vNV == nullptr ||
glMakeBufferResidentNV == nullptr) {
console::warn("Unable to initialize solution '%s', GL_NV_shader_buffer_load unavailable.", GetName().c_str());
return false;
}
if (!UntexturedObjectsSolution::Init(_vertices, _indices, _objectCount)) {
return false;
}
// Program
const char* kUniformNames[] = { "ViewProjection", "World", nullptr };
m_prog = CreateProgramT("cubes_gl_bindless_vs.glsl",
"cubes_gl_bindless_fs.glsl",
kUniformNames, &mUniformLocation);
if (m_prog == 0) {
console::warn("Unable to initialize solution '%s', shader compilation/linking failed.", GetName().c_str());
return false;
}
m_ibs.resize(_objectCount);
m_ib_addrs.resize(_objectCount);
m_ib_sizes.resize(_objectCount);
m_vbs.resize(_objectCount);
m_vbo_addrs.resize(_objectCount);
m_vbo_sizes.resize(_objectCount);
glGenBuffers(_objectCount, &*m_ibs.begin());
glGenBuffers(_objectCount, &*m_vbs.begin());
for (size_t u = 0; u < _objectCount; ++u)
{
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_ibs[u]);
glBufferStorage(GL_ELEMENT_ARRAY_BUFFER, _indices.size() * sizeof(UntexturedObjectsProblem::Index), &*_indices.begin(), 0);
glGetBufferParameterui64vNV(GL_ELEMENT_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_ib_addrs[u]);
glMakeBufferResidentNV(GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY);
m_ib_sizes[u] = _indices.size() * sizeof(UntexturedObjectsProblem::Index);
glBindBuffer(GL_ARRAY_BUFFER, m_vbs[u]);
glBufferStorage(GL_ARRAY_BUFFER, _vertices.size() * sizeof(UntexturedObjectsProblem::Vertex), &*_vertices.begin(), 0);
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_vbo_addrs[u]);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
m_vbo_sizes[u] = _vertices.size() * sizeof(UntexturedObjectsProblem::Vertex);
}
return glGetError() == GL_NO_ERROR;
}
EXTERN_C_ENTER
JNIEXPORT void JNICALL Java_org_lwjgl_opengl_NVShaderBufferLoad_glMakeBufferResidentNV(JNIEnv *__env, jclass clazz, jint target, jint access) {
glMakeBufferResidentNVPROC glMakeBufferResidentNV = (glMakeBufferResidentNVPROC)tlsGetFunction(2099);
UNUSED_PARAM(clazz)
glMakeBufferResidentNV(target, access);
}
bool initArrayBuffer()
{
glGenBuffers(1, &BufferName);
glBindBuffer(GL_ARRAY_BUFFER, BufferName);
glBufferData(GL_ARRAY_BUFFER, VertexSize, VertexData, GL_STATIC_DRAW);
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &Address);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
glBindBuffer(GL_ARRAY_BUFFER, 0);
return glf::checkError("initArrayBuffer");;
}
bool initBuffer()
{
glGenBuffers(buffer::MAX, BufferName);
glBindBuffer(GL_ARRAY_BUFFER, BufferName[buffer::VERTEX]);
glBufferData(GL_ARRAY_BUFFER, VertexSize, VertexData, GL_STATIC_DRAW);
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &Address);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
glBindBuffer(GL_ARRAY_BUFFER, 0);
glBindBuffer(GL_UNIFORM_BUFFER, BufferName[buffer::TRANSFORM]);
glBufferData(GL_UNIFORM_BUFFER, sizeof(glm::mat4), 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_UNIFORM_BUFFER, 0);
return true;
}
int voxel_zread(int bricks, FILE * fptr)
{
unsigned char in[CHUNK];
unsigned char out[CHUNK];
z_stream strm = {
.zalloc=Z_NULL,
.zfree=Z_NULL,
.opaque=Z_NULL,
.avail_in=0,
.next_in=Z_NULL
};
int ret = inflateInit(&strm);
if(ret != Z_OK)
return ret;
int prev_off = 0;
int brick_size = B_SIZE*B_SIZE*B_SIZE*sizeof(float4);
unsigned char *prev_buf = malloc(brick_size);
do {
strm.avail_in = fread(in, 1, CHUNK, fptr);
if(ferror(fptr)) {
ret = Z_ERRNO;
goto zread_fail;
}
if(strm.avail_in == 0)
break;
strm.next_in = in;
do {
strm.avail_out = CHUNK;
strm.next_out = out;
ret = inflate(&strm, Z_NO_FLUSH);
// assert(ret != Z_STREAM_ERROR);
switch(ret) {
case Z_NEED_DICT:
ret = Z_DATA_ERROR;
case Z_DATA_ERROR:
case Z_MEM_ERROR:
goto zread_fail;
}
int have = CHUNK - strm.avail_out;
// pass data to loader
int outoff = 0;
if(have+prev_off > brick_size)
{
if(prev_off)
{
outoff = brick_size - prev_off;
memcpy(prev_buf+prev_off, out, outoff);
bp_add((float4*)prev_buf);
prev_off = 0;
}
for(;outoff+brick_size <= have; outoff+= brick_size)
bp_add((float4*)(out+outoff));
}
if(outoff < have)
{
memcpy(prev_buf+prev_off, out+outoff, have - outoff);
prev_off += have - outoff;
if(prev_off >= brick_size)
{
bp_add((float4*)prev_buf);
prev_off = 0;
}
}
// strm.avail_out = 0;
// end data loader
} while(strm.avail_out == 0);
} while(ret != Z_STREAM_END);
inflateEnd(&strm);
printf("zlib ok: inflated %d bricks.\n", brick_count-1);
return 0;
zread_fail:
inflateEnd(&strm);
free(prev_buf);
printf("zlib fail = %d, bricks=%d\n", ret, brick_count);
return ret;
}
int voxel_init(char *filename)
{
bPool = malloc(bPool_size);
memset(bPool, 0, bPool_size);
nPool = malloc(nPool_size);
memset(nPool, 0, nPool_size);
// Load file off of disk
FILE *fptr = fopen(filename, "rb");
if(!fptr)
{
printf("Failed Opening SVO file \"%s\"\n", filename);
return 0;
}
char check[4] = "VOCT";
int bricks, blocks;
int zero;
fread(check, 4, 1, fptr);
fread(&blocks, 4, 1, fptr);
fread(&bricks, 4, 1, fptr);
fread(&zero, 4, 1, fptr);
fread(nPool, blocks * 64, 1, fptr);
int ret = voxel_zread(bricks, fptr);
if(ret)
{
printf("zload failed\n");
}
/*
for(int i=0; i < bricks; i++)
{
int3 boff;
int itmp = i % B_COUNT;
boff.x = itmp;
itmp = (i-boff.x)/B_COUNT;
boff.y = itmp % B_COUNT;
boff.z = (itmp-boff.y) / B_COUNT;
F3MULS(boff, boff, B_SIZE);
for(int z=0; z < B_SIZE; z++)
for(int y=0; y < B_SIZE; y++)
{
fread(&bPool[boff.z+z][boff.y+y][boff.x],
sizeof(float4)*B_SIZE, 1, fptr);
}
}
*/
// fread(bPool, 16*B_CUBE, 1, fptr);
fclose(fptr);
printf("Loaded %d blocks, %d bricks\n", blocks, bricks);
printf("Error = \"%s\"\n", glError(glGetError()));
free_vram();
// Send the brick pool to the GPU
glEnable(GL_TEXTURE_3D);
glGenTextures(1, &brick_tex);
glBindTexture(GL_TEXTURE_3D, brick_tex);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_S, GL_CLAMP);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_T, GL_CLAMP);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_R, GL_CLAMP);
printf("Error = \"%s\"\n", glError(glGetError()));
glTexImage3D(GL_TEXTURE_3D, 0, GL_RGBA16F,
B_EDGE, B_EDGE, B_EDGE,
0, GL_RGBA, GL_FLOAT, bPool);
printf("Error = \"%s\"\n", glError(glGetError()));
glBindTexture(GL_TEXTURE_3D, 0);
printf("Error = \"%s\"\n", glError(glGetError()));
free_vram();
// load the node pool onto the GPU
glGenBuffers(1, &node_pool);
glBindBuffer(GL_ARRAY_BUFFER, node_pool);
glBufferData(GL_ARRAY_BUFFER, nPool_size, nPool, GL_DYNAMIC_DRAW);
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER,
GL_BUFFER_GPU_ADDRESS_NV, &nodepool_devptr);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
glBindBuffer(GL_ARRAY_BUFFER, 0);
printf("Error = \"%s\"\n", glError(glGetError()));
prepare_shader();
return 1;
}
void GLGeometry::refresh() {
Assert(m_id[0] != 0 && m_id[1] != 0);
m_stride = 3;
if (m_mesh->hasVertexNormals())
m_stride += 3;
if (m_mesh->hasVertexTexcoords())
m_stride += 2;
if (m_mesh->hasUVTangents())
m_stride += 3;
if (m_mesh->hasVertexColors())
m_stride += 3;
m_stride *= sizeof(GLfloat);
size_t vertexCount = m_mesh->getVertexCount(), triCount = m_mesh->getTriangleCount();
m_size[EVertexID] = (GLuint) (vertexCount * m_stride);
m_size[EIndexID] = (GLuint) (triCount * sizeof(GLuint) * 3);
Log(ETrace, "Uploading a GPU geometry object (\"%s\", " SIZE_T_FMT
" vertices, " SIZE_T_FMT " triangles, %s)",
getName().c_str(), vertexCount, triCount,
memString(m_size[EVertexID] + m_size[EIndexID]).c_str());
GLfloat *vertices = new GLfloat[vertexCount * m_stride/sizeof(GLfloat)];
GLuint *indices = (GLuint *) m_mesh->getTriangles();
const Point *sourcePositions = m_mesh->getVertexPositions();
const Normal *sourceNormals = m_mesh->getVertexNormals();
const Point2 *sourceTexcoords = m_mesh->getVertexTexcoords();
const Color3 *sourceColors = m_mesh->getVertexColors();
Vector *sourceTangents = NULL;
if (m_mesh->hasUVTangents()) {
/* Convert into per-vertex tangents */
const TangentSpace *triTangents = m_mesh->getUVTangents();
sourceTangents = new Vector[vertexCount];
uint32_t *count = new uint32_t[vertexCount];
memset(sourceTangents, 0, sizeof(Vector)*vertexCount);
for (size_t i=0; i<triCount; ++i) {
const Triangle &tri = m_mesh->getTriangles()[i];
const TangentSpace &tangents = triTangents[i];
for (int j=0; j<3; ++j) {
sourceTangents[tri.idx[j]] += tangents.dpdu;
++count[tri.idx[j]];
}
}
for (size_t i=0; i<vertexCount; ++i) {
if (count[i] == 0)
continue;
sourceTangents[i] /= (Float) count[i];
}
delete[] count;
}
size_t pos = 0;
for (size_t i=0; i<vertexCount; ++i) {
vertices[pos++] = (GLfloat) sourcePositions[i].x;
vertices[pos++] = (GLfloat) sourcePositions[i].y;
vertices[pos++] = (GLfloat) sourcePositions[i].z;
if (sourceNormals) {
vertices[pos++] = (GLfloat) sourceNormals[i].x;
vertices[pos++] = (GLfloat) sourceNormals[i].y;
vertices[pos++] = (GLfloat) sourceNormals[i].z;
}
if (sourceTexcoords) {
vertices[pos++] = (GLfloat) sourceTexcoords[i].x;
vertices[pos++] = (GLfloat) sourceTexcoords[i].y;
}
if (sourceTangents) {
vertices[pos++] = (GLfloat) sourceTangents[i].x;
vertices[pos++] = (GLfloat) sourceTangents[i].y;
vertices[pos++] = (GLfloat) sourceTangents[i].z;
}
if (sourceColors) {
vertices[pos++] = (GLfloat) sourceColors[i][0];
vertices[pos++] = (GLfloat) sourceColors[i][1];
vertices[pos++] = (GLfloat) sourceColors[i][2];
}
}
Assert(pos * sizeof(GLfloat) == m_stride * vertexCount);
bind();
glBufferData(GL_ARRAY_BUFFER, m_size[EVertexID], vertices, GL_STATIC_DRAW);
if (GLEW_NV_vertex_buffer_unified_memory) {
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_addr[EVertexID]);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
}
glBufferData(GL_ELEMENT_ARRAY_BUFFER, m_size[EIndexID], indices, GL_STATIC_DRAW);
if (GLEW_NV_vertex_buffer_unified_memory) {
glGetBufferParameterui64vNV(GL_ELEMENT_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_addr[EIndexID]);
glMakeBufferResidentNV(GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY);
}
unbind();
delete[] vertices;
if (sourceTangents)
delete[] sourceTangents;
}
// --------------------------------------------------------------------------------------------------------------------
bool UntexturedObjectsGLBindlessIndirect::Init(const std::vector<UntexturedObjectsProblem::Vertex>& _vertices,
const std::vector<UntexturedObjectsProblem::Index>& _indices,
size_t _objectCount)
{
if (glBufferStorage == nullptr) {
console::warn("Unable to initialize solution '%s', glBufferStorage() unavailable.", GetName().c_str());
return false;
}
if (glGetBufferParameterui64vNV == nullptr ||
glMakeBufferResidentNV == nullptr) {
console::warn("Unable to initialize solution '%s', GL_NV_shader_buffer_load unavailable.", GetName().c_str());
return false;
}
if (!UntexturedObjectsSolution::Init(_vertices, _indices, _objectCount)) {
return false;
}
// Program
const char* kUniformNames[] = { "ViewProjection", "World", nullptr };
m_prog = CreateProgramT("cubes_gl_bindless_indirect_vs.glsl",
"cubes_gl_bindless_indirect_fs.glsl",
kUniformNames, &mUniformLocation);
if (m_prog == 0) {
console::warn("Unable to initialize solution '%s', shader compilation/linking failed.", GetName().c_str());
return false;
}
m_ibs.resize(_objectCount);
m_ib_addrs.resize(_objectCount);
m_ib_sizes.resize(_objectCount);
m_vbs.resize(_objectCount);
m_vbo_addrs.resize(_objectCount);
m_vbo_sizes.resize(_objectCount);
m_commands.resize(_objectCount);
glGenBuffers(_objectCount, &*m_ibs.begin());
glGenBuffers(_objectCount, &*m_vbs.begin());
for (size_t u = 0; u < _objectCount; ++u)
{
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_ibs[u]);
glBufferStorage(GL_ELEMENT_ARRAY_BUFFER, _indices.size() * sizeof(UntexturedObjectsProblem::Index), &*_indices.begin(), 0);
glGetBufferParameterui64vNV(GL_ELEMENT_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_ib_addrs[u]);
glMakeBufferResidentNV(GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY);
m_ib_sizes[u] = _indices.size() * sizeof(UntexturedObjectsProblem::Index);
glBindBuffer(GL_ARRAY_BUFFER, m_vbs[u]);
glBufferStorage(GL_ARRAY_BUFFER, _vertices.size() * sizeof(UntexturedObjectsProblem::Vertex), &*_vertices.begin(), 0);
glGetBufferParameterui64vNV(GL_ARRAY_BUFFER, GL_BUFFER_GPU_ADDRESS_NV, &m_vbo_addrs[u]);
glMakeBufferResidentNV(GL_ARRAY_BUFFER, GL_READ_ONLY);
m_vbo_sizes[u] = _vertices.size() * sizeof(UntexturedObjectsProblem::Vertex);
}
glGenBuffers(1, &m_transform_buffer);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, m_transform_buffer);
glBufferStorage(GL_SHADER_STORAGE_BUFFER, _objectCount * 64, nullptr, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_DYNAMIC_STORAGE_BIT);
m_transform_ptr = glMapBufferRange(GL_SHADER_STORAGE_BUFFER, 0, _objectCount * 64, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT);
glGenBuffers(1, &m_cmd_buffer);
glBindBuffer(GL_DRAW_INDIRECT_BUFFER, m_cmd_buffer);
glBufferStorage(GL_DRAW_INDIRECT_BUFFER, _objectCount * sizeof(Command), nullptr, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_DYNAMIC_STORAGE_BIT);
m_cmd_ptr = glMapBufferRange(GL_DRAW_INDIRECT_BUFFER, 0, _objectCount * sizeof(Command), GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT);
m_queries.resize(4);
glGenQueries(kQueryCount, &*m_queries.begin());
return glGetError() == GL_NO_ERROR;
}