Merge branch 'master' into blender2.8

This commit is contained in:
Luca Rood 2017-07-05 15:50:01 +02:00
commit bdeeb29482
11 changed files with 120 additions and 42 deletions

View File

@ -149,7 +149,8 @@ public:
device_memory& use_queues_flag,
device_memory& work_pool_wgs);
virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&);
virtual SplitKernelFunction* get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures&);
virtual int2 split_kernel_local_size();
virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task);
virtual uint64_t state_buffer_size(device_memory& kg, device_memory& data, size_t num_threads);
@ -932,7 +933,8 @@ bool CPUSplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim,
return true;
}
SplitKernelFunction* CPUSplitKernel::get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&)
SplitKernelFunction* CPUSplitKernel::get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures&)
{
CPUSplitKernelFunction *kernel = new CPUSplitKernelFunction(device);

View File

@ -105,7 +105,8 @@ public:
device_memory& use_queues_flag,
device_memory& work_pool_wgs);
virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&);
virtual SplitKernelFunction* get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures&);
virtual int2 split_kernel_local_size();
virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task);
};
@ -2058,7 +2059,8 @@ bool CUDASplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim
return !device->have_error();
}
SplitKernelFunction* CUDASplitKernel::get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&)
SplitKernelFunction* CUDASplitKernel::get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures&)
{
CUfunction func;

View File

@ -125,7 +125,8 @@ public:
device_memory& use_queues_flag,
device_memory& work_pool_wgs) = 0;
virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&) = 0;
virtual SplitKernelFunction* get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures&) = 0;
virtual int2 split_kernel_local_size() = 0;
virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task) = 0;
};

View File

@ -84,7 +84,7 @@ public:
string *error = NULL);
static bool device_version_check(cl_device_id device,
string *error = NULL);
static string get_hardware_id(string platform_name,
static string get_hardware_id(const string& platform_name,
cl_device_id device_id);
static void get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices,
bool force_all = false);
@ -247,17 +247,17 @@ public:
public:
OpenCLProgram() : loaded(false), device(NULL) {}
OpenCLProgram(OpenCLDeviceBase *device,
string program_name,
string kernel_name,
string kernel_build_options,
const string& program_name,
const string& kernel_name,
const string& kernel_build_options,
bool use_stdout = true);
~OpenCLProgram();
void add_kernel(ustring name);
void load();
bool is_loaded() { return loaded; }
string get_log() { return log; }
bool is_loaded() const { return loaded; }
const string& get_log() const { return log; }
void report_error();
cl_kernel operator()();
@ -271,8 +271,8 @@ public:
bool load_binary(const string& clbin, const string *debug_src = NULL);
bool save_binary(const string& clbin);
void add_log(string msg, bool is_debug);
void add_error(string msg);
void add_log(const string& msg, bool is_debug);
void add_error(const string& msg);
bool loaded;
cl_program program;

View File

@ -263,7 +263,7 @@ public:
explicit OpenCLSplitKernel(OpenCLDeviceSplitKernel *device) : DeviceSplitKernel(device), device(device) {
}
virtual SplitKernelFunction* get_split_kernel_function(string kernel_name,
virtual SplitKernelFunction* get_split_kernel_function(const string& kernel_name,
const DeviceRequestedFeatures& requested_features)
{
OpenCLSplitKernelFunction* kernel = new OpenCLSplitKernelFunction(device, cached_memory);

View File

@ -241,9 +241,9 @@ string OpenCLCache::get_kernel_md5()
}
OpenCLDeviceBase::OpenCLProgram::OpenCLProgram(OpenCLDeviceBase *device,
string program_name,
string kernel_file,
string kernel_build_options,
const string& program_name,
const string& kernel_file,
const string& kernel_build_options,
bool use_stdout)
: device(device),
program_name(program_name),
@ -274,7 +274,7 @@ void OpenCLDeviceBase::OpenCLProgram::release()
}
}
void OpenCLDeviceBase::OpenCLProgram::add_log(string msg, bool debug)
void OpenCLDeviceBase::OpenCLProgram::add_log(const string& msg, bool debug)
{
if(!use_stdout) {
log += msg + "\n";
@ -288,7 +288,7 @@ void OpenCLDeviceBase::OpenCLProgram::add_log(string msg, bool debug)
}
}
void OpenCLDeviceBase::OpenCLProgram::add_error(string msg)
void OpenCLDeviceBase::OpenCLProgram::add_error(const string& msg)
{
if(use_stdout) {
fprintf(stderr, "%s\n", msg.c_str());
@ -707,7 +707,7 @@ bool OpenCLInfo::device_version_check(cl_device_id device,
return true;
}
string OpenCLInfo::get_hardware_id(string platform_name, cl_device_id device_id)
string OpenCLInfo::get_hardware_id(const string& platform_name, cl_device_id device_id)
{
if(platform_name == "AMD Accelerated Parallel Processing" || platform_name == "Apple") {
/* Use cl_amd_device_topology extension. */

View File

@ -19,28 +19,28 @@
#if defined(WITH_CYCLES_LOGGING) && !defined(__KERNEL_GPU__)
# include <glog/logging.h>
#else
# include <iostream>
#endif
CCL_NAMESPACE_BEGIN
#if !defined(WITH_CYCLES_LOGGING) || defined(__KERNEL_GPU__)
class StubStream : public std::ostream {
public:
StubStream() : std::ostream(NULL) { }
class StubStream {
public:
template<class T>
StubStream& operator<<(const T&) {
return *this;
}
};
class LogMessageVoidify {
public:
LogMessageVoidify() { }
void operator&(::std::ostream&) { }
void operator&(StubStream&) { }
};
# define LOG_SUPPRESS() (true) ? (void) 0 : LogMessageVoidify() & StubStream()
# define LOG(severity) LOG_SUPPRESS()
# define VLOG(severity) LOG_SUPPRESS()
#endif
#define VLOG_ONCE(level, flag) if(!flag) flag = true, VLOG(level)

View File

@ -3379,7 +3379,8 @@ void BKE_particlesettings_make_local(Main *bmain, ParticleSettings *part, const
/* Textures */
/************************************************/
static int get_particle_uv(DerivedMesh *dm, ParticleData *pa, int face_index, const float fuv[4], char *name, float *texco)
static int get_particle_uv(DerivedMesh *dm, ParticleData *pa, int index, const float fuv[4],
char *name, float *texco, bool from_vert)
{
MFace *mf;
MTFace *tf;
@ -3395,11 +3396,15 @@ static int get_particle_uv(DerivedMesh *dm, ParticleData *pa, int face_index, co
if (pa) {
i = ELEM(pa->num_dmcache, DMCACHE_NOTFOUND, DMCACHE_ISCHILD) ? pa->num : pa->num_dmcache;
if (i >= dm->getNumTessFaces(dm))
if ((!from_vert && i >= dm->getNumTessFaces(dm)) ||
(from_vert && i >= dm->getNumVerts(dm)))
{
i = -1;
}
}
else {
i = index;
}
else
i = face_index;
if (i == -1) {
texco[0] = 0.0f;
@ -3407,7 +3412,19 @@ static int get_particle_uv(DerivedMesh *dm, ParticleData *pa, int face_index, co
texco[2] = 0.0f;
}
else {
mf = dm->getTessFaceData(dm, i, CD_MFACE);
if (from_vert) {
mf = dm->getTessFaceDataArray(dm, CD_MFACE);
for (int j = 0; j < dm->getNumTessFaces(dm); j++, mf++) {
if (ELEM(i, mf->v1, mf->v2, mf->v3, mf->v4)) {
i = j;
break;
}
}
}
else {
mf = dm->getTessFaceData(dm, i, CD_MFACE);
}
psys_interpolate_uvs(&tf[i], mf->v4, fuv, texco);
@ -3474,8 +3491,11 @@ static void get_cpa_texture(DerivedMesh *dm, ParticleSystem *psys, ParticleSetti
mul_m4_v3(mtex->object->imat, texvec);
break;
case TEXCO_UV:
if (fw && get_particle_uv(dm, NULL, face_index, fw, mtex->uvname, texvec))
if (fw && get_particle_uv(dm, NULL, face_index, fw, mtex->uvname,
texvec, (part->from == PART_FROM_VERT)))
{
break;
}
/* no break, failed to get uv's, so let's try orco's */
ATTR_FALLTHROUGH;
case TEXCO_ORCO:
@ -3547,8 +3567,11 @@ void psys_get_texture(ParticleSimulationData *sim, ParticleData *pa, ParticleTex
mul_m4_v3(mtex->object->imat, texvec);
break;
case TEXCO_UV:
if (get_particle_uv(sim->psmd->dm_final, pa, 0, pa->fuv, mtex->uvname, texvec))
if (get_particle_uv(sim->psmd->dm_final, pa, 0, pa->fuv, mtex->uvname,
texvec, (part->from == PART_FROM_VERT)))
{
break;
}
/* no break, failed to get uv's, so let's try orco's */
ATTR_FALLTHROUGH;
case TEXCO_ORCO:

View File

@ -427,12 +427,34 @@ static int distribute_binary_search(float *sum, int n, float value)
static void distribute_from_verts_exec(ParticleTask *thread, ParticleData *pa, int p)
{
ParticleThreadContext *ctx= thread->ctx;
int rng_skip_tot= PSYS_RND_DIST_SKIP; /* count how many rng_* calls wont need skipping */
MFace *mface;
DM_ensure_tessface(ctx->dm);
mface = ctx->dm->getTessFaceDataArray(ctx->dm, CD_MFACE);
int rng_skip_tot = PSYS_RND_DIST_SKIP; /* count how many rng_* calls wont need skipping */
/* TODO_PARTICLE - use original index */
pa->num= ctx->index[p];
pa->fuv[0] = 1.0f;
pa->fuv[1] = pa->fuv[2] = pa->fuv[3] = 0.0;
pa->num = ctx->index[p];
zero_v4(pa->fuv);
if (pa->num != DMCACHE_NOTFOUND && pa->num < ctx->dm->getNumVerts(ctx->dm)) {
for (int i = 0; i < ctx->dm->getNumTessFaces(ctx->dm); i++, mface++) {
if (ELEM(pa->num, mface->v1, mface->v2, mface->v3, mface->v4)) {
unsigned int *vert = &mface->v1;
for (int j = 0; j < 4; j++, vert++) {
if (*vert == pa->num) {
pa->fuv[j] = 1.0f;
break;
}
}
break;
}
}
}
#if ONLY_WORKING_WITH_PA_VERTS
if (ctx->tree) {

View File

@ -286,9 +286,9 @@ void InverseSearchRadiusOperation::initExecution()
this->m_inputRadius = this->getInputSocketReader(0);
}
voi *InverseSearchRadiusOperation::initializeTileData(rcti *rect)
void *InverseSearchRadiusOperation::initializeTileData(rcti *rect)
{
MemoryBuffer * data = new MemoryBuffer(NULL, rect);
MemoryBuffer * data = new MemoryBuffer(COM_DT_COLOR, rect);
float *buffer = data->getBuffer();
int x, y;
int width = this->m_inputRadius->getWidth();
@ -343,7 +343,7 @@ voi *InverseSearchRadiusOperation::initializeTileData(rcti *rect)
void InverseSearchRadiusOperation::executePixelChunk(float output[4], int x, int y, void *data)
{
MemoryBuffer *buffer = (MemoryBuffer *)data;
buffer->readNoCheck(color, x, y);
buffer->readNoCheck(output, x, y);
}
void InverseSearchRadiusOperation::deinitializeTileData(rcti *rect, void *data)

View File

@ -448,10 +448,12 @@ static int rna_ParticleSystem_tessfaceidx_on_emitter(ParticleSystem *particlesys
int totpart;
int totchild = 0;
int totface;
int totvert;
int num = -1;
DM_ensure_tessface(modifier->dm_final); /* BMESH - UNTIL MODIFIER IS UPDATED FOR MPoly */
totface = modifier->dm_final->getNumTessFaces(modifier->dm_final);
totvert = modifier->dm_final->getNumVerts(modifier->dm_final);
/* 1. check that everything is ok & updated */
if (!particlesystem || !totface) {
@ -486,13 +488,26 @@ static int rna_ParticleSystem_tessfaceidx_on_emitter(ParticleSystem *particlesys
return num;
}
}
else if (part->from == PART_FROM_VERT) {
if (num != DMCACHE_NOTFOUND && num < totvert) {
MFace *mface = modifier->dm_final->getTessFaceDataArray(modifier->dm_final, CD_MFACE);
*r_fuv = &particle->fuv;
for (int i = 0; i < totface; i++, mface++) {
if (ELEM(num, mface->v1, mface->v2, mface->v3, mface->v4)) {
return i;
}
}
}
}
}
else {
ChildParticle *cpa = particlesystem->child + particle_no - totpart;
num = cpa->num;
if (part->childtype == PART_CHILD_FACES) {
if (ELEM(part->from, PART_FROM_FACE, PART_FROM_VOLUME)) {
if (ELEM(part->from, PART_FROM_FACE, PART_FROM_VOLUME, PART_FROM_VERT)) {
if (num != DMCACHE_NOTFOUND && num < totface) {
*r_fuv = &cpa->fuv;
return num;
@ -512,6 +527,19 @@ static int rna_ParticleSystem_tessfaceidx_on_emitter(ParticleSystem *particlesys
return num;
}
}
else if (part->from == PART_FROM_VERT) {
if (num != DMCACHE_NOTFOUND && num < totvert) {
MFace *mface = modifier->dm_final->getTessFaceDataArray(modifier->dm_final, CD_MFACE);
*r_fuv = &parent->fuv;
for (int i = 0; i < totface; i++, mface++) {
if (ELEM(num, mface->v1, mface->v2, mface->v3, mface->v4)) {
return i;
}
}
}
}
}
}